fix(sip): bound RTP allocation to configured range

[ryanmurf]

Summary

• load Asterisk-compatible [general] rtpstart/rtpend settings from rtp.conf and fail startup on invalid configured ranges
• reserve UDP sockets through one bounded allocator shared by inbound and outbound SIP call paths
• return SIP 503 and remove transient call state when the range is exhausted
• release the reservation with channel teardown and update the inbound media E2E to assert the advertised port is bounded

Asterisk documents rtpstart/rtpend as the inclusive RTP range in its canonical rtp.conf sample: https://github.com/asterisk/asterisk/blob/master/configs/samples/rtp.conf.sample

The SDP answer continues to advertise the actual bound UDP port in the media description, per RFC 4566 section 5.14: https://www.rfc-editor.org/rfc/rfc4566#section-5.14

Regression coverage

• parses and validates Asterisk-style RTP range settings
• exhausts a one-port allocator, then reuses the port after the session drops
• exhausts a one-port call range, then reuses it after ChannelDriver::hangup tears down the first call
• verifies inbound exhaustion returns 100 then 503 with no live handler or driver call state
• verifies the authenticated inbound RTP echo path advertises a port inside the default bounded range

Validation

• cargo +1.97.0 clippy --workspace --exclude pjsip-shim -- -D warnings
• cargo +1.97.0 test --workspace --exclude pjsip-shim

Fixes #70