Optimize SampleBuilder and reset cross-sample state on Flush

[solos]

Description

Improve SampleBuilder performance and make Flush() a clean boundary for a new RTP stream.
Performance:

• Replace lastSampleTimestamp *uint32 with uint32 + hasLastSampleTimestamp to avoid a heap allocation on every emitted sample
• Pre-size sample payload buffers from the total RTP payload length
• Compute Sample.Duration with integer arithmetic instead of float64 Flush semantics:
• After purging buffered packets, Flush() now clears lastSampleTimestamp, droppedPackets, and paddingPackets
• Prevents post-flush packets from being misclassified as padding from a prior stream when timestamps collide
• Prevents PrevDroppedPackets on the next sample from including drops that occurred during the flush of leftover buffer data Samples built during Flush() still capture PrevDroppedPackets at build time; only state for packets received after Flush() is reset. Tests:
• Add TestSampleBuilder_FlushResetsState
• Add TestSampleBuilder_ValidSampleSameTimestampAfterFlush

Reference issue

None

[codecov]

Codecov Report

❌ Patch coverage is 82.35294% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.46%. Comparing base (a323cae) to head (9b109df).

Files with missing lines	Patch %	Lines
pkg/media/samplebuilder/samplebuilder.go	82.35%	1 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3445      +/-   ##
==========================================
- Coverage   85.60%   85.46%   -0.14%     
==========================================
  Files          81       81              
  Lines        9856     9868      +12     
==========================================
- Hits         8437     8434       -3     
- Misses       1003     1011       +8     
- Partials      416      423       +7     

Flag	Coverage Δ
go	85.46% <82.35%> (-0.14%)	⬇️

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

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[JoTurk]

Change looks good to me, but we'll have to manually test it because samplebuilder has many moving parts.

[JoTurk]

The old behavior returned invalid value (+Inf) if the sample-rate is 0.00f, this will just panic, both aren't ideal, but if there is a setup where users can pass sample rate to applications, this can be used to trigger panic/DoS.

Maybe we can check and return an error early if the user created a samplebuilder with 0 sample rate?

[solos]

Yes, you are right. If we return an error, the caller will need to change too. What about set the Duration to 0.