Fix creation of streams with RST_FRAME.

[juliens]

This PR fixes a performance issue + timeouts.

After the end of the stream (remove from the stream map) sometimes some RST_STREAM can happen and then recreate streams. As RST_STREAM should never happen first on a stream, ignoring those frame for the creation fixes the problem.

Before:

❯ hey -c 100 -z 30s -h2 -t 5 https://127.0.0.1:8443

Summary:
  Total:    30.3452 secs
  Slowest:  0.0182 secs
  Fastest:  0.0001 secs
  Average:  0.0026 secs
  Requests/sec: 430.7106

  Total data:   124700 bytes
  Size/request: 10 bytes

Response time histogram:
  0.000 [1] |
  0.002 [2595]  |■■■■■■■■■■■■
  0.004 [8440]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.006 [1212]  |■■■■■■
  0.007 [153]   |■
  0.009 [25]    |
  0.011 [11]    |
  0.013 [8] |
  0.015 [12]    |
  0.016 [8] |
  0.018 [5] |

Latency distribution:
  10% in 0.0013 secs
  25% in 0.0020 secs
  50% in 0.0026 secs
  75% in 0.0031 secs
  90% in 0.0038 secs
  95% in 0.0044 secs
  99% in 0.0063 secs

Details (average, fastest, slowest):
  DNS+dialup:   0.0000 secs, 0.0001 secs, 0.0182 secs
  DNS-lookup:   0.0000 secs, 0.0000 secs, 0.0000 secs
  req write:    0.0000 secs, 0.0000 secs, 0.0043 secs
  resp wait:    0.0008 secs, 0.0000 secs, 0.0041 secs
  resp read:    0.0002 secs, 0.0000 secs, 0.0041 secs

Status code distribution:
  [200] 12470 responses

Error distribution:
  [87]  Get "https://127.0.0.1:8443": context deadline exceeded
  [513] Get "https://127.0.0.1:8443": context deadline exceeded (Client.Timeout exceeded while awaiting headers)

After

❯ hey -c 100 -z 30s -h2 -t 5 https://127.0.0.1:8443

Summary:
  Total:    30.0029 secs
  Slowest:  0.0283 secs
  Fastest:  0.0000 secs
  Average:  0.0030 secs
  Requests/sec: 40354.8588

  Total data:   12107620 bytes
  Size/request: 12 bytes

Response time histogram:
  0.000 [1] |
  0.003 [676267]    |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.006 [321576]    |■■■■■■■■■■■■■■■■■■■
  0.009 [2023]  |
  0.011 [34]    |
  0.014 [17]    |
  0.017 [23]    |
  0.020 [24]    |
  0.023 [15]    |
  0.025 [5] |
  0.028 [15]    |

Latency distribution:
  10% in 0.0012 secs
  25% in 0.0020 secs
  50% in 0.0025 secs
  75% in 0.0030 secs
  90% in 0.0035 secs
  95% in 0.0039 secs
  99% in 0.0048 secs

Details (average, fastest, slowest):
  DNS+dialup:   0.0000 secs, 0.0000 secs, 0.0283 secs
  DNS-lookup:   0.0000 secs, 0.0000 secs, 0.0000 secs
  req write:    0.0000 secs, 0.0000 secs, 0.0051 secs
  resp wait:    0.0009 secs, 0.0000 secs, 0.0139 secs
  resp read:    0.0002 secs, 0.0000 secs, 0.0067 secs

Status code distribution:
  [200] 1000000 responses

[dgrr]

Thanks for your PR!