Memory regression tests

[alan-george-lk]

This PR introduces a new test binary called memory_tester for Linux, which has the following features:

• Configurable common scenarios
  • Examples: publishing tracks, room joining/closure
• RSS and thread usage tracking
• Configurable repeated iterations, acceptable memory and thread growth bounds

CI is also updated to run this tester and provide a nice markdown table report of the results, and fail jobs that introduce changes that fail this.

Documentation is also updated around testing in general, as it was a bit out of date.

Results

Here are two FFI release results with the following params:

Iterations: 100
Warmup: 20
Threshold: RSS growth ≤ 8 MiB; thread growth ≤ 0.

Prior release livekit-ffi/v0.12.81 results:

Scenario	RSS 0→20	RSS 20→100	Threads 0→20	Threads 20→100	Status
audio-track	+60.63 MiB	-2.64 MiB	+28	0	✅ PASS
video-track	+61.43 MiB	-2.62 MiB	+28	0	✅ PASS
data-track	+53.88 MiB	+0.03 MiB	+2	0	✅ PASS
room-reuse	+53.54 MiB	+3.47 MiB	+2	0	✅ PASS
room-client-leave	+53.66 MiB	+0.12 MiB	+2	0	✅ PASS
room-server-delete	+143.46 MiB	+108.94 MiB	+105	+400	❌ FAIL

Current FFI release results:

Scenario	RSS 0→20	RSS 20→100	Threads 0→20	Threads 20→100	Status
audio-track	+17.36 MiB	+0.19 MiB	+12	0	✅ PASS
video-track	+16.74 MiB	-0.05 MiB	+7	0	✅ PASS
data-track	+12.30 MiB	+0.03 MiB	+1	0	✅ PASS
room-reuse	+12.09 MiB	0.00 MiB	+1	0	✅ PASS
room-client-leave	+12.10 MiB	+0.05 MiB	+1	0	✅ PASS
room-server-delete	+12.11 MiB	+0.08 MiB	+1	0	✅ PASS

[devin-ai-integration]

Note

Newer findings are available below. Devin Review posted a newer report on this PR, in addition to the findings presented here.

🔍 Devin Review: 2 flags

Not posted on this PR by your GitHub settings — view them in Devin Review. (Configure)