refactor(ext/http): use scopeguard defer to handle async drop

[lrowe]

Use the scopeguard defer macro to run cleanup code for new_slab_future.
This means it can be a single async function, avoiding the need to create a struct and implement PinnedDrop

Async cleanup in Rust is awkward because async functions may be cancelled at any await point when their Future is dropped.
The scopeguard approach comes from the following articles:

• How to think about async/await in Rust
• Async Cancellation I (Reddit discussion)

[lrowe]

I assume the earlier windows test failure were spurious since it passed when running in my fork here: https://github.com/lrowe/deno/actions/runs/6287030911

[lrowe]

This failure is also spurious: error: Import 'https://deno.land/x/[email protected]/mod.ts' failed: 502 Bad Gateway

[mmastrac]

LGTM

[lrowe]

Benchmark results seem to show no change in performance.

This branch. ./deno-b129b8950 run --allow-all multi.ts minimal

% third_party/prebuilt/mac/wrk http:https://localhost:8080/
Running 10s test @ http:https://localhost:8080/
  2 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    77.17us   22.10us 634.00us   73.23%
    Req/Sec    57.49k     1.58k   59.19k    92.57%
  1155607 requests in 10.10s, 165.31MB read
Requests/sec: 114417.97
Transfer/sec:     16.37MB

Main. ./deno-cb9ab9c3a run --allow-all multi.ts minimal

% third_party/prebuilt/mac/wrk http:https://localhost:8080/
Running 10s test @ http:https://localhost:8080/
  2 threads and 10 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency    77.25us   22.15us 676.00us   72.89%
    Req/Sec    57.39k     1.35k   59.16k    90.10%
  1153767 requests in 10.10s, 165.05MB read
Requests/sec: 114235.08
Transfer/sec:     16.34MB

[lrowe]

Given that the windows tests seem to consistently fail the same tests on this branch when using the xl windows runner I'm going to try to get GitHub actions to run just the failing tests to see and add logging to see what's going on.

[serve_test 020.85] httpServerTcpCancellation_file ... FAILED (107ms)
[serve_test 020.90] httpServerTcpCancellation_stream ... FAILED (44ms)

EDIT: The failing Test debug (fast) step does not use the xl runner, so I don't know what the difference is between the test running on my fork vs here.

[mmastrac]

@lrowe I think Luca may have fixed the flaky tests in #20672

[lrowe]

Great! I've added that commit to this PR for now and will update after it is merged.