Remote Debug

[jrcheli]

This can be thought of as an epic that potentially has a few very different, separable sub-tasks:

1. the library logging output is currently "just a bunch of unparseable text". I think it'd be a welcome improvement to wrap these in a json message, with fields like pid, severity, proc name, timestamp that are distinct from the text of the message.
2. Right now the SCOPE_LOG_LEVEL isn't fine-grained enough to be able to look at specific problems. We're in a situation where we either have too much debugging or not enough. Consider adding flags to enable/disable categories of logs, maybe? For example, maybe one category for "wrap.c what interposed functions are getting called" another for "hooking/attach", maybe another for "protocol debugging", etc.
3. When fatal signals happen (segfaults, for example), we may want to add a "debugging capture" thing to our library so can capture pertinent info (config files, memory, stack) ala a core-dump. Disabled by default? Enabled by a env var, maybe? IDK.
4. We currently only have an arcane way of getting debug counters (the results of DBG() statements) out of scoped application... "Dynamic configuration" with SCOPE_CMD_DBG_PATH... There is a fair amount that could be helpful: switch this output to json, create this output on a remote command too ("GetDiag").
5. Add a self-diagnostic capability to scope? cli or ldscope option?

scope self-test
# Using /etc/scope/scope.yml
## testing the TCP transport path
### Events Received
### Metrics Received
### Logs Received

[seanvaleo]

Superceded by #1279 and PR #1289