fix(ext/http): Use brotli compression params

[mmastrac]

Fixes #19737 by adding brotli compression parameters.

Time after:

Accept-Encoding: gzip:

real    0m0.214s
user    0m0.005s
sys     0m0.013s

Accept-Encoding: br:

Before:

real    0m10.303s
user    0m0.005s
sys     0m0.010s

After:

real    0m0.127s
user    0m0.006s
sys     0m0.014s

[bartlomieju]

LGTM, is there a test that could be added to avoid regressions in the future?

[mmastrac]

LGTM, is there a test that could be added to avoid regressions in the future?

We might be able to create a timing comparison test between GZip and Brotli but I worry it might be flaky. Maybe a heuristic like "does this compress within 10x the wall-clock time of GZip", but that could be a challenge to find a good set of parameters that detect the problem and are sensitive enough to trigger if this code gets lost. I wouldn't recommend using a fixed wall-clock timeout for this because that will almost certainly trigger too often.

I think I should punt on this one because it's probably a bit too much work.

[rajsite]

@mmastrac know how to check if this change is on Deno Deploy yet? I think it may be the root cause of denoland/deploy_feedback#434

(Also not sure if you need to redeploy for updates, etc)

[lucab]

@rajsite I answered directly there. This fix was not on Deno Deploy at the time, but it is applied there too right now.