feat: Add support for NO_PROXY for websocket connections

[jbw-vtl]

In #16326 the team added support for HTTP_PROXY & HTTPS_PROXY for the websocket connection used by the agent to submit events (such as task runs).
However in some environments these proxies are necessary to route connection for the flow itself running (i.e. to connect to external in an air gapped environment).
Having the events routed through the proxy might mean they cannot reach the server.

This PR adds basic support for the NO_PROXY variable for the websocket connection, allowing to ignore the proxy for the server connection.

Checklist

• This pull request references any related issue by including "closes <link to issue>"
  • If no issue exists and your change is not a small fix, please create an issue first.
• If this pull request adds new functionality, it includes unit tests that cover the changes
• [ ] If this pull request removes docs files, it includes redirect settings in mint.json.
• If this pull request adds functions or classes, it includes helpful docstrings.

This stemmed from the discussion around #15153 and closes #16537

[codspeed-hq]

CodSpeed Performance Report

Merging #16538 will not alter performance

_{Comparing jbw-vtl:main (195aba2) with main (441f721)}

Summary

✅ 3 untouched benchmarks

[jbw-vtl]

Adjusted the used proxy to be http as even a HTTPS_PROXY needs to be http, the tests without mocking were failing with the below

def parse_proxy_url(url: str) -> Tuple[ProxyType, str, int, Optional[str], Optional[str]]:
        parsed = urlparse(url)
    
        scheme = parsed.scheme
        if scheme == 'socks5':
            proxy_type = ProxyType.SOCKS5
        elif scheme == 'socks4':
            proxy_type = ProxyType.SOCKS4
        elif scheme == 'http':
            proxy_type = ProxyType.HTTP
        else:
>           raise ValueError(f'Invalid scheme component: {scheme}')  # pragma: no cover
E           ValueError: Invalid scheme component: https

[jakekaplan]

I'm inclined to use from urllib.request import proxy_bypass here to avoid having our own implementation if possible.

My reading of NO_PROXY is that it's either an exact match or substring (as opposed to just a suffix), which would conflict with this as written but would be supported by proxy_bypass.

I'm not super familiar, did you see something that suggested it should work differently?

[jbw-vtl]

Was looking for precisely this, also not super familiar here.
This looks indeed perfect, swapped it out and working as expected.
Pushed a change to swap out just now

[jakekaplan]

@jbw-vtl will need a review from a core maintainer to merge but this LGTM. Thanks for contributing the fix!

[zzstoatzz]

thank you @jbw-vtl !