fix: Explicit mtSQUELCH classification in Message::compress switch (#6740)

[svirdi-ripple]

High Level Overview of Change

Fix for #6740: Message::compress() in src/xrpld/overlay/detail/Message.cpp enumerates message types in a switch statement to decide compressibility, but mtSQUELCH was absent from both the compressible and non-compressible groups. Silent fall-through happened to produce the correct behavior (uncompressed), but any future MessageType enum value would silently inherit the same fall-through — a maintenance hazard flagged by the NOLINTNEXTLINE(bugprone-switch-missing-default-case) suppression.

This PR:

• Adds case protocol::mtSQUELCH: break; to the non-compressible group.
• Adds a default: case with XRPL_ASSERT(false, ...) so future omissions surface loudly in debug/test builds instead of silently falling through.
• Removes the now-redundant NOLINTNEXTLINE suppression.
• Adds a regression test in src/test/overlay/compression_test.cpp (testMtSquelchUncompressed) that constructs a TMSquelch payload padded above the 70-byte compression gate and asserts getBuffer(Compressed::On) and getBuffer(Compressed::Off) return byte-identical buffers — locking in the non-compressible classification.

Context of Change

INFORMATIONAL — no security impact, no observable behavior change. Code hygiene / maintenance fix.

Continues work from closed PR #6805, which had the code change but no test, kept the NOLINTNEXTLINE, and was closed due to unsigned commits.

API Impact

• Public API: New feature (new methods and/or new fields)
• Public API: Breaking change
• libxrpl change
• Peer protocol change

Internal xrpld/overlay change only — no public or wire-protocol impact.

Test plan

Run the compression suite:

```
./xrpld --unittest=xrpl.overlay.compression --unittest-log
```

• Full suite passes with the fix in place (16 cases, 0 failures).
• New `testMtSquelchUncompressed` case runs and asserts the compressed/uncompressed buffers are byte-identical for a padded `TMSquelch` payload.