Fix test_cli and test_cookies

[ChihweiLHBird]

I was able to reproduce the test cli issues on my laptop, which is the timeout occurred when running the subprocess in capture function. After increasing the timeout locally, the issue never came up again during multiple tests.

Open this PR to further test the test cases.

I think this issue usually happen on smaller machine with lower performance.

[codecov]

Codecov Report

Merging #2479 (6da4c06) into main (ce926a3) will not change coverage.
The diff coverage is n/a.

@@            Coverage Diff            @@
##              main     #2479   +/-   ##
=========================================
  Coverage   87.267%   87.267%           
=========================================
  Files           60        60           
  Lines         5097      5097           
  Branches       913       913           
=========================================
  Hits          4448      4448           
  Misses         476       476           
  Partials       173       173           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ce926a3...6da4c06. Read the comment docs.

[ChihweiLHBird]

Test cookies issue seems a problem of httpx's async client.
encode/httpx#2144

In the failed test cookie case, the headers contain setting cookie but hpptx didn't put it into response.cookies. I think there is not much we can do until httpx dev team fix it. If we really want to fix it right now, we can try to parse the cookies in sanic-testing after getting the response object from httpx.

[ChihweiLHBird]

It's weird that fixing cli tests appears increase the chance of the failed cookies test. @ahopkins

[ChihweiLHBird]

The cookies test fixed, the only (occasionally) broken test case is keep alive.

[ahopkins]

Amazing amazing work. You rock.