fix: Don't handle fork in SWOOLE_BASE mode

[PROFeNoM]

Description

Trying to set the callbacks before the instance is created is senseless: That's the bug.

--> This PR defers the dummy callbacks set to the posthook, so we have an instance.

What's more, swoole_fork_exec is not necessarily called. It only happens when the server isn't in "base" mode. The default behavior of swoole is to be in base mode, hence explaining why the tests were failing when the hook was done correctly (because no fork was happening - i.e., the pid was the same all along).

Why are the Laravel Octane tests working? Because it is, by default, using the SWOOLE_PROCESS mode.

--> With this PR, the fork is only handled when it happens - i.e., when is_base_mode evaluates to false.

Reviewer checklist

• Test coverage seems ok.
• Appropriate labels assigned.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0% with 5 lines in your changes are missing coverage. Please review.

Project coverage is 79.20%. Comparing base (8388bc1) to head (154869e).
Report is 4 commits behind head on master.

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #2656      +/-   ##
============================================
+ Coverage     77.66%   79.20%   +1.54%     
- Complexity     2205     2207       +2     
============================================
  Files           225      199      -26     
  Lines         26019    22007    -4012     
  Branches        988        0     -988     
============================================
- Hits          20208    17431    -2777     
+ Misses         5285     4576     -709     
+ Partials        526        0     -526     

Flag	Coverage Δ
appsec-extension	?
tracer-extension	78.47% <ø> (ø)
tracer-php	80.23% <0.00%> (-0.05%)	⬇️

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

Files	Coverage Δ
.../DDTrace/Integrations/Swoole/SwooleIntegration.php	0.00% <0.00%> (ø)

... and 26 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 8388bc1...154869e. Read the comment docs.

[pr-commenter]

Benchmarks

Benchmark execution time: 2024-05-03 13:04:37

Comparing candidate commit 154869e in PR branch alex/fix/swoole-callback with baseline commit 8388bc1 in branch master.

Found 2 performance improvements and 0 performance regressions! Performance is the same for 176 metrics, 0 unstable metrics.

scenario:ContextPropagationBench/benchInject64Bit-opcache

• 🟩 execution_time [-478.290ns; -165.710ns] or [-6.538%; -2.265%]

scenario:TraceSerializationBench/benchSerializeTrace-opcache

• 🟩 mem_peak [-155.258KB; -51.287KB] or [-7.000%; -2.312%]

[labbati]

There was already a test failing to prove that this fix actually fixes the issue?

[PROFeNoM]

@labbati Tests were failing when the hook was changed from a pre-hook to a post-hook. Traces were not received because we were supposing a fork would happen, while none did, hence introducing an issue with the background sender.