Fix DTMF deduplication to use (timestamp, event code) composite key

[cuihang]

Problem

The DTMF handler in dtmfHandler (media_port.go) deduplicated incoming DTMF packets using only the RTP timestamp:

if h.Timestamp == p.lastDTMFTimestamp.Load() {
    return nil
}

RFC 4733 requires all packets of a given digit to share identical timestamps, so this correctly filters redundant packets within one digit. However, some SIP devices or carriers reuse the timestamp of the previous digit when sending the next digit. When this happens, the next digit is incorrectly filtered out and never reported.

Fix

Replace the timestamp-only dedup with a (timestamp, event code) composite key:

ev, err := dtmf.Decode(payload)
if err != nil {
    return nil
}
eventID := uint64(h.Timestamp)<<8 | uint64(ev.Code)
if eventID == p.lastDTMFEvent.Load() {
    return nil
}
p.lastDTMFEvent.Store(eventID)
fnc(ev)

Behavior

Scenario	Before	After
Same digit, redundant packets (same ts + same code)	Deduped ✓	Deduped ✓
Different digits sharing timestamp (same ts + different code, e.g. *, 0, 1)	Incorrectly dropped ✗	Each reported ✓
No dependency on RTP marker bit	✓	✓

Known limitation

If an upstream sends two identical digits (e.g. two 0s) with the exact same timestamp and event code, the (timestamp, event code) key alone cannot distinguish them. Resolving that case requires full DTMF lifecycle state tracking (marker bit, End bit, duration reset, sequence number gaps). That is a more complex enhancement beyond this fix; in the worst case (all fields identical) the receiver cannot distinguish the digits at all.

Changes

• pkg/sip/media_port.go: Replace lastDTMFTimestamp atomic.Uint32 with lastDTMFEvent atomic.Uint64; update dtmfHandler to decode first, then dedup on composite key.
• pkg/sip/media_port_test.go: Update field reference; add TestMediaPortDTMFSameTimestamp verifying two different digits (0 and 1) sharing the same RTP timestamp are both reported.

Test plan

• TestMediaPortDTMF — all 12 existing subtests pass (digits 1/12/123 × loss none/first/last/middle)
• TestMediaPortDTMFSameTimestamp — new test passes: digits 0 and 1 with same timestamp → both reported as 01

[CLAassistant]

All committers have signed the CLA.

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 1 additional finding.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 66.69%. Comparing base (0460b40) to head (135a5fc).
⚠️ Report is 348 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #796      +/-   ##
==========================================
+ Coverage   65.25%   66.69%   +1.44%     
==========================================
  Files          51       41      -10     
  Lines        6588     8096    +1508     
==========================================
+ Hits         4299     5400    +1101     
- Misses       1915     2209     +294     
- Partials      374      487     +113     

☔ 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.