Do not check get_DD_TRACE_ENABLED() outside of a request

[bwoebi]

get_DD_TRACE_ENABLED() must not be accessed outside of requests, given that it accesses a request local array of configs.

DDTRACE_G(active_stack) is supposed to be always properly reset between requests, so we can just put this first and then the calls to get_DD_TRACE_ENABLED() are always correct.

Fixes a crash in datadog_profiling::timeline::timeline_mshutdown doing collect_idle() and ultimately accessing the profiling_context here.

Not sure how to test this; it was reported with the pretty obvious backtrace by an user.

[codecov-commenter]

Codecov Report

Merging #2631 (e6caca9) into master (8180fdb) will decrease coverage by 2.18%.
The diff coverage is n/a.

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #2631      +/-   ##
============================================
- Coverage     78.67%   76.49%   -2.18%     
  Complexity     2198     2198              
============================================
  Files           199      130      -69     
  Lines         21955    13095    -8860     
  Branches          0      986     +986     
============================================
- Hits          17273    10017    -7256     
+ Misses         4682     2552    -2130     
- Partials          0      526     +526     

Flag	Coverage Δ
appsec-extension	69.00% <ø> (?)
tracer-extension	?
tracer-php	79.79% <ø> (+0.90%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 132 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8180fdb...e6caca9. Read the comment docs.

[pr-commenter]

Benchmarks

Benchmark execution time: 2024-04-17 17:28:04

Comparing candidate commit e6caca9 in PR branch bob/fix-profiler-mshutdown-crash with baseline commit 8180fdb in branch master.

Found 1 performance improvements and 1 performance regressions! Performance is the same for 176 metrics, 0 unstable metrics.

scenario:ContextPropagationBench/benchInject64Bit

• 🟩 execution_time [-461.407ns; -178.593ns] or [-6.298%; -2.438%]

scenario:TraceSerializationBench/benchSerializeTrace

• 🟥 execution_time [+4.310µs; +6.990µs] or [+2.382%; +3.863%]

[morrisonlevi]

Thank you for this PR. We should definitely write down some of this knowledge on conditional checking order in a code comment so it is not forgotten.

[morrisonlevi]

Does it make any sense to fetch DDTRACE_G(active_stack) && DDTRACE_G(active_stack)->root_span from mshutdown? I think we may need to avoid any call to this function after post-request shutdown or so.

[bwoebi]

These globals are valid until gshutdown actually, so it's fine.