fix: remove implicit Content-Type magic in res.set (#7145)

[luvsharma6342]

What’s included in the PR

lib/response.js: removed the Content-Type “magic” from res.set (no implicit MIME lookup / no charset injection); updated JSDoc
test/res.set.js: added assertions for pass-through Content-Type
test/res.send.js: updated expectations for Buffer/Uint8Array cases that previously depended on the old res.set behavior
Tests I ran locally
npx mocha ... test/res.set.js (passing)
npx mocha ... test/res.send.js (passing)

[notadev99]

This is a behavior-breaking change — today res.set('Content-Type', 'text/plain') yields text/plain; charset=utf-8 and 'html' resolves to text/html; charset=utf-8; after this both pass through verbatim. A few things worth nailing down:

Release target. Which line is this aimed at? Dropping the implicit charset changes real responses, so it reads like major-version material rather than a patch.

Migration path / blast radius. The JSDoc now points to res.type() for charset and shorthand, which is the right pointer. Can you confirm the common res.send('...') path (no explicit Content-Type) still emits charset=utf-8? The two res.send test edits suggest the default may now drop charset as well, which would be a much wider change than just explicit res.set calls.

Compat note. Serving text/html with no charset lets a browser sniff the encoding. Modern browsers default to UTF-8 so this is mostly historical, but a line in the changelog/migration notes would help anyone relying on the implicit charset know to switch to res.type().

Coverage for the new pass-through behavior looks good.