Replace logging implementation with spdlog

[msimberg]

This replaces the internal (boost logging-based) logging implementation with spdlog (https://github.com/gabime/spdlog). At the same time I've noticeably simplified the logging setup.

Compared to before, there is only one log now which can be used with PIKA_LOG(level, "msg {}", arg). Logging is done to stderr by default. The default log level is warnings and more severe. The default logging output looks like this now (a snippet from hello_world):

[2024-04-16 17:27:38.942414484] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180341] [pool:0000/0003/0003] [parent:----/----] [task:0x7f01ec028b00/----] [runtime.cpp:400/set_state] runtime_state::startup
[2024-04-16 17:27:38.942424521] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180341] [pool:0000/0003/0003] [parent:----/----] [task:0x7f01ec028b00/----] [runtime.cpp:1023/run_helper] (4th stage) run_helper: ran startup functions
[2024-04-16 17:27:38.942433060] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180341] [pool:0000/0003/0003] [parent:----/----] [task:0x7f01ec028b00/----] [runtime.cpp:1026/run_helper] (4th stage) runtime::run_helper: bootstrap complete
[2024-04-16 17:27:38.942440892] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180341] [pool:0000/0003/0003] [parent:----/----] [task:0x7f01ec028b00/----] [runtime.cpp:400/set_state] runtime_state::running
[2024-04-16 17:27:38.942450837] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180341] [pool:0000/0003/0003] [parent:----/----] [task:0x7f01ec028b00/----] [runtime.cpp:1032/run_helper] (last stage) runtime::run_helper: about to invoke pika_main
Hello World!
[2024-04-16 17:27:38.942485701] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180336] [pool:----/0004/----] [parent:----/----] [task:----/----] [runtime.cpp:1189/wait] runtime: about to enter wait state
[2024-04-16 17:27:38.942594025] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180342] [pool:----/----/----] [parent:----/----] [task:----/----] [runtime.cpp:1158/wait_finalize] runtime: about to enter wait state
[2024-04-16 17:27:38.942607966] [pika] [info] [host:pf3dx1(-1)] [pid:2180336] [tid:2180342] [pool:----/----/----] [parent:----/----] [task:----/----] [runtime.cpp:1160/wait_finalize] runtime: exiting wait state

• The time and pika is always printed first (useful to users who might not know that pika is running under the hood, at least they have something to search for)
• The log level comes next, and is printed in color (see color: https://github.com/gabime/spdlog/wiki/3.-Custom-formatting#pattern-flags)
• The hostname is printed along with the MPI rank (if defined), based on

  pika/libs/pika/debugging/include/pika/debugging/print.hpp

  Lines 280 to 287 in adb63f0

  	struct hostname_print_helper
  	{
  	PIKA_EXPORT char const* get_hostname() const;
  	PIKA_EXPORT int guess_rank() const;
  	PIKA_EXPORT friend std::ostream& operator<<(
  	std::ostream& os, hostname_print_helper const& h);
  	};

• The process and thread id are printed with the formatting provided by spdlog (%P and %t from https://github.com/gabime/spdlog/wiki/3.-Custom-formatting#pattern-flags)
• The pool and worker thread are printed in the format <pool num>/<global thread num>/<local thread num>. ---- is printed when any of them is not set.
• The parent and self task ids are printed with the default thread_id formatter.
• The filename, line number, and function are always printed to add some context.
• The actual log message is always printed last.

The log format is customizable with PIKA_LOG_FORMAT or --pika:log-format. The pika information (thread ids, hostname, worker thread information) is implemented using custom spdlog formatters.

The log level follows spdlog's log levels, from 0 (trace) to 6 (off) and can be set with PIKA_LOG_LEVEL or --pika:log-level.

The destination can be set with PIKA_LOG_DESTINATION or --pika:log-destination, and defaults to cerr. This can be cerr or cout for stderr and stdout, or a file otherwise.

[pika-bot]

Performance test report

pika Performance

Comparison

BENCHMARK	RESULT
Task Overhead - Create Thread Hierarchical - Latch	-

Info

Property	Before	After
pika Commit	0abc084	1b8bbb6
pika Datetime	2024-02-19T15:15:15+00:00	2024-05-03T17:09:08+00:00
Envfile
Compiler	/apps/daint/SSL/pika/spack/lib/spack/env/clang/clang++ 11.0.1	/apps/daint/SSL/pika/spack/lib/spack/env/clang/clang++ 11.0.1
Hostname	nid00025	nid00025
Clustername	daint	daint
Datetime	2024-02-19T16:26:16.072067+01:00	2024-05-03T19:17:05.041455+02:00

Explanation of Symbols

Symbol	MEANING
=	No performance change (confidence interval within ±1%)
(=)	Probably no performance change (confidence interval within ±2%)
(+)/(-)	Very small performance improvement/degradation (≤1%)
+/-	Small performance improvement/degradation (>10%)
++/--	Large performance improvement/degradation (>10%)
+++/---	Very large performance improvement/degradation (>10%)
?	Probably no change, but quite large uncertainty (confidence interval with ±5%)
??	Unclear result, very large uncertainty (±10%)
???	Something unexpected…

[pika-bot]

Performance test report

pika Performance

Comparison

BENCHMARK	RESULT
Task Overhead - Create Thread Hierarchical - Latch	-

Info

Property	Before	After
pika Commit	0abc084	058ebd8
pika Datetime	2024-02-19T15:15:15+00:00	2024-05-06T11:57:16+00:00
Envfile
Datetime	2024-02-19T16:26:16.072067+01:00	2024-05-06T14:11:42.348357+02:00
Clustername	daint	daint
Compiler	/apps/daint/SSL/pika/spack/lib/spack/env/clang/clang++ 11.0.1	/apps/daint/SSL/pika/spack/lib/spack/env/clang/clang++ 11.0.1
Hostname	nid00025	nid00025

Explanation of Symbols

Symbol	MEANING
=	No performance change (confidence interval within ±1%)
(=)	Probably no performance change (confidence interval within ±2%)
(+)/(-)	Very small performance improvement/degradation (≤1%)
+/-	Small performance improvement/degradation (>10%)
++/--	Large performance improvement/degradation (>10%)
+++/---	Very large performance improvement/degradation (>10%)
?	Probably no change, but quite large uncertainty (confidence interval with ±5%)
??	Unclear result, very large uncertainty (±10%)
???	Something unexpected…