Add WatchStreamExt::reflect to allow chaining on a reflector

[clux]

Thought this might be nicer than the separate function call now that we allow doing everything on the stream chain. It also makes it clearer to see what parts get reflected (particularly pre or post modify calls) because it follows from the chain order:

watcher(api, watcher::Config::default().any_semantic())
    .default_backoff()
    .modify(|pod| { pod.managed_fields_mut().clear(); })
    .reflect(writer)
    .applied_objects();

NB: I did some rename experiments on this branch but it was a bad idea and force pushed it away. Sorry about the noise.

[codecov]

Codecov Report

Merging #1252 (562b6c1) into main (6c9a531) will increase coverage by 0.13%.
The diff coverage is 97.05%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1252      +/-   ##
==========================================
+ Coverage   73.05%   73.19%   +0.13%     
==========================================
  Files          74       75       +1     
  Lines        5991     6025      +34     
==========================================
+ Hits         4377     4410      +33     
- Misses       1614     1615       +1     

Impacted Files	Coverage Δ
kube-runtime/src/controller/mod.rs	35.15% <ø> (ø)
kube-runtime/src/utils/mod.rs	62.96% <ø> (ø)
kube-runtime/src/utils/watch_ext.rs	11.76% <0.00%> (-0.74%)	⬇️
kube-runtime/src/utils/reflect.rs	100.00% <100.00%> (ø)

[Dav1dde]

Just some small nits.

Should the kube::runtime::reflector be removed in favor of this?

[clux]

Just some small nits.

Appreciate it!

Should the kube::runtime::reflector be removed in favor of this?

Maybe. I wouldn't want to remove the fn immediately on the introduction here, but if this doesn't have any downsides, then we can probably put a longer lasting deprecation marker on reflector() in the future. For now have only tried to make some reasonable set of docs on the new method and changed over some internal usage.