Hotfix duplicate share checker

A bug was discovered and patched in which some duplicate shares would not be detected

Minor bugs and improvements, sphinx docs

Bugs

• default SHA256 hashes_per_share config was wrong
• Dont record shares in the share chain for merge mined currencies (thanks @sbwdlihao)

Improvements

• StratumServer now holds recent jobs in memory so that stale share tracking is more accurate. Also allows stale share tracking currency/algorithm information.
• Live job storage moved from Jobmanager to the StratumServer, allowing a more event like system and reducing coupling between Jobmanager and StratumServer

Docs

• Basic documentation has been added to ReadTheDocs, migrating and updating the README
• Configuration example has more helpful information about configuration

Bug fixes

• SIGHUP is now handled properly performing cleanup for all components
• job_mapper dictionary gets cleaned up every 10 pushes (keeping at least 10 pushes work of job mappings for stale difficulty determination)
• Masternodes (Darkcoin) are now paid out dynamically only when needed
• A Nagios style check script added to contrib for checking stratum availability and performance

Addition of statsd support and several usability improvements

Bug Fixes

• "Status" keys from ppagent now expire properly
• Graceful shutdown now flushes queued "one minute shares" properly

Features

• Optional "events" config value sends various helpful performance stats to a statsd endpoint over UDP
• Blocking detector logs traceback when gevent hasn't context switched in 1 second
• Monitor views are substantially expanded to include more information
• Mining client version is now recorded on client object
• Added a minimum manual share difficulty setting
• Pool shares can now be reported in several very flexible ways
• All port numbers will now be based off of a "server_number" which is now a command line option as well, making clustering much easier

Improve abstractions through big rework

A bit rewrite again that improves decoupling between components in preparation for running them in separate processes.

Major rewrite with tons of improvements

Breaking Changes

• Now requires version 0.2.3 of cryptokit!
• Configuration file format has completely changed. Work from the config.yml.example file again.
• Requires very latest pre-release v0.7 of SimpleCoin. Optionally see commits simplecrypto/simplecoin@51c45b8, simplecrypto/simplecoin@06cc7dd, and simplecrypto/simplecoin@446d8b6 for suport

Features

• Big rewrites across codebase to a more class based, modular design.
• Reporter and jobmanager are now modular, allowing different modules to be used by simply changing config options.
• Default CeleryReporter now aggregates statistics and shares at a server wide level (instead of on a per-client basis) greatly reducing database chatter. For instance, 200 workers connected using the same address and worker name will only commit changes to the database twice a minute.
• Binding multiple stratum ports and agent server ports in the same server is now supported. Each port can have different configurations (vardiff on/off, different fixed diff, etc)
• Configurable idle worker detection
• Several new stats are now being collected, including number of celery tasks sent, new counts from every aux chain, and many more.
• Highly optimized job pushing code, reducing stale share amounts by 30-60% over previous version
• Greatly enhanced monitoring output. Logs number of solved/rejected/accepted blocks for both primary and aux chains, among several other useful bits of information.
• Daemon now gracefully shutsdown when HUP is sent, allowing upstart's reload command to work as desired without losing any aggregated shares.
• Timing for statistics collection is now much lower overhead and much more precise.
• Full push block notifications for both aux chains and main chains is now supported through the use of OS signals.

Changes

• All blocks are now submitted by the jobmanager, isolating all RPC server interation to the jobmanager.
• Jobs now determine the algorithm that should be used, allowing more flexibility in the jobmanager.
• Several profiling entries have been added allowing detection of poor performance sections for future optimization.

Major rewrite with tons of improvements

Breaking Changes

• Now requires version 0.2.3 of cryptokit!
• Configuration file format has completely changed. Work from the config.yml.example file again.
• Requires very latest pre-release v0.7 of SimpleCoin. Optionally see commits simplecrypto/simplecoin@51c45b8, simplecrypto/simplecoin@06cc7dd, and simplecrypto/simplecoin@446d8b6 for suport

Features

• Big rewrites across codebase to a more class based, modular design.
• Reporter and jobmanager are now modular, allowing different modules to be used by simply changing config options.
• Default CeleryReporter now aggregates statistics and shares at a server wide level (instead of on a per-client basis) greatly reducing database chatter. For instance, 200 workers connected using the same address and worker name will only commit changes to the database twice a minute.
• Binding multiple stratum ports and agent server ports in the same server is now supported. Each port can have different configurations (vardiff on/off, different fixed diff, etc)
• Configurable idle worker detection
• Several new stats are now being collected, including number of celery tasks sent, new counts from every aux chain, and many more.
• Highly optimized job pushing code, reducing stale share amounts by 30-60% over previous version
• Greatly enhanced monitoring output. Logs number of solved/rejected/accepted blocks for both primary and aux chains, among several other useful bits of information.
• Daemon now gracefully shutsdown when HUP is sent, allowing upstart's reload command to work as desired without losing any aggregated shares.
• Timing for statistics collection is now much lower overhead and much more precise.
• Full push block notifications for both aux chains and main chains is now supported through the use of OS signals.

Changes

• All blocks are now submitted by the jobmanager, isolating all RPC server interation to the jobmanager.
• Jobs now determine the algorithm that should be used, allowing more flexibility in the jobmanager.
• Several profiling entries have been added allowing detection of poor performance sections for future optimization.

Share overreport fix

Features

• Stale/Dup/Low percentages are now shown at main powerpool stats endpoint for convenience
• Hashrate on /clients stats endpoint is now correct
• process title for powerpool can now be set with new configuration value procname. Make sure you install the new requirements!

Bugs

• Shares were being overreported because of the timing jitter optimization being slightly mis-implemented. Overreporting was random, so shouldn't effect payout fairness, but does inflate share counts and hashrates...

Transaction inclusion

Features

• More information about recently solved blocks is available through the web monitoring port

Bugs

• Transactions are now included in blocks after resolution of missing/spent input bug.

Merged mining and share batching

Breaking Changes

• Now requires version 0.2 of cryptokit!

Features

• All shares are batched into one minute chunks instead of logging each individual share. Huge decrease in IO load and rows in SQL
• Complete support for merge mining an arbitrary number of currencies. Allows configurable pushing and flushing of jobs for new aux work.

Bugs

• A few more properly closing file descriptor problems resolved.