remove whitespace in transfer-encoding

[pajod]

HTTP parser Whitespace cleanup

1. one no-op replacement
2. one variable reuse after casting to tuple, to clarify these are still just SP and HTAB
3. another no-op replacement
4. a meaningful change for Gunicorn incorrectly strips certain characters from transfer-codings #3364
   • N.B. this still disallows Content-Encoding: identity, (trailing comma), probably a robust choice
   • no longer meaningful after 2073e13 - \x0b is rejected by the new RFC9110_5_5_INVALID_AND_DANGEROUS
   • N.B. responding 501 is not what the original reporter recommended - but differentiating invalid and unsupported is is complicated because of potential parameters, and I do no see how the difference matters.
5. a no-op replacement in checking the CONNECTION header.. which should probably just be dropped. How could v != v.strip(" \t") legitimately happen, unless something was wrong with obsolete folding?

Searching for strip() revealed additional references in cfg.strip_header_spaces (known dangerous) in util.is_hoppish() (unreachable after 72238fc - left as-is but did add the extra header rfc9112 tells us to drop) and in the websockets example (not reviewed yet).

References:

• https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.3
• https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.1.2

[pajod]

I don't see any legitimate use for this. I suggest Gunicorn removes the .strip() call.