fix: detect lr parameter with value as loose routing in ACK

[bevenky]

Problem

Some SIP proxies (e.g., Kamailio) send Record-Route with ;lr=on instead of bare ;lr:

Record-Route: <sip:proxy.example.com;lr=on;ftag=abc;did=xyz>

The rsip crate parses ;lr=on as Param::Other("lr", Some("on")) since Param::Lr only matches the valueless form (;lr with no =value).

This causes make_ack to fall into the strict routing branch instead of loose routing:

• The ACK Request-URI is set to the Record-Route URI
• The Contact (often an internal IP like 10.x.x.x) is placed in the Route header
• destination_from_request picks the Route header as the UDP destination
• The ACK is sent to the unreachable internal IP instead of through the proxy
• The remote side never receives the ACK → "ACK Timeout" after ~30 seconds → call killed

This is similar to the issue fixed in PR #100 (incorrect Request-URI for ACK), but for the lr parameter detection specifically.

Fix

Check for both Param::Lr and Param::Other where the name is "lr" (case-insensitive):

.any(|param| matches!(param, rsip::Param::Lr) 
    || matches!(param, rsip::Param::Other(name, _) if name.to_string().eq_ignore_ascii_case("lr")))

This matches ;lr, ;lr=on, ;lr=true, etc. as loose routing per RFC 3261 Section 16.6.

Testing

Tested with a SIP proxy that sends Record-Route: <sip:proxy.example.com;lr=on;...>:

• Before fix: ACK sent to internal Contact IP (e.g. 10.x.x.x:5060), call dies after 30s
• After fix: ACK routed through Record-Route proxy, call stays alive

Note

The root cause is in the rsip crate's Param parser which only matches bare ;lr as Param::Lr (line 119 in rsip/src/common/uri/param/mod.rs). A fix there would also resolve this, but the workaround in make_ack is simpler and handles any lr variant.

[yeoleobun]

Thanks