feat: Use memoized request_class and response_class values

[kedod]

Description

• Use memoized request_class and response_class values

Closes

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.23%. Comparing base (d02ecbb) to head (2f9206e).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3205   +/-   ##
========================================
  Coverage    98.23%   98.23%           
========================================
  Files          321      321           
  Lines        14482    14488    +6     
  Branches      2302     2304    +2     
========================================
+ Hits         14227    14233    +6     
  Misses         113      113           
  Partials       142      142           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[cofin]

LGTM

[cofin]

@all-contributors add @kedod for code

[allcontributors]

@cofin

@kedod already contributed before to code

[peterschutt]

Can we do resolve_response_class() at the same time? Its documented as being memoized just like this one, but isn't.

[peterschutt]

For interests sake, the memoization of resolve_response_class() was removed in #357.

That PR added resolve_response_handler() which is itself cached and appears to be the only place that we call resolve_response_class() internally - so I can understand why its caching was removed in that case.

However, it has since always been documented as being cached, even though it isn't, and we offer it as public API, so if users are calling it more than once they'd benefit from the caching as documented and it wouldn't really cost us anything. So, now that we are caching resolve_request_class() I think its it probably better to memoize it too so the similar methods are consistent.

In that PR we also removed caching for resolve_response_cookies() and resolve_response_headers(), but the docstrings of those methods were updated to reflect that.

[kedod]

@peterschutt I discovered exactly the same thing last evening and removed the changes for the resolve_response_class.
But being consistent in the public APIs is a valid point.
I've added memoization on the resolve_response_class level.

[github-actions]

Documentation preview will be available shortly at https://litestar-org.github.io/litestar-docs-preview/3205