Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@test/http/compression.e2e.spec.ts`:
- Around line 37-41: The test for non-compressible content-type is ineffective
because image(`@Res`() res: UwsResponse) currently sends a tiny Buffer so
compression is skipped by the threshold rather than by content-type; update the
test handler (image) to send a payload larger than the compression threshold
(e.g., allocate >1024 bytes) or set the compression threshold to 0 for this test
so compression decision is driven by Content-Type 'image/png' and then keep the
assertion block in the spec (the checks around compression headers/assertions)
to validate that image/png is not compressed.
- Around line 60-61: Replace the fixed port constant (port = 13358) and any
other hardcoded port usage with dynamic port allocation and bind to the loopback
address: stop using the literal port variable and instead start the test server
with port 0 and host '127.0.0.1' (e.g., server.listen(0, '127.0.0.1', ...)),
then read the assigned port from server.address().port for constructing URLs;
update any references to the original port variable (including the other
occurrence around line 81) to use the discovered port so tests avoid port
collisions and use an explicit loopback host.
- Around line 105-117: The httpGet helper currently creates an
http.ClientRequest (variable req) but never sets a request-level timeout, so
tests can hang; update the httpGet function to call req.setTimeout(<timeoutMs>,
() => req.destroy(new Error('Request timeout'))) (or similar) after creating
req, ensure the timeout value is configurable or a defined constant, and make
sure the request 'error' handler and response listeners properly handle the
destroyed request so the Promise rejects on timeout; reference the existing
local variable req and the httpGet function to locate where to add
req.setTimeout and the timeout error handling.