fast retransmit fix

[JPDye]

Stacked PR. Requires #1156 and CUBIC. Will rebase once that merges.

Previously, the fast retransmit would send the entire buffer. Now, it only sends the oldest segment. On the single flow netsim, performance regresses and on the multiflow netsim (#1153) performance is significantly improved. This is because fast-retransmit works better with "realistic" tail drop loss better than it does random loss.

These performance losses are somewhat regained by changing the RTTEstimator to handle RTO differently to individual retransmissions, but still leave the single flow reasonably worse off when random loss occurs.

Single flow:

╭───┬───────┬───────┬───────┬───────┬────────┬────────┬────────┬────────┬────────╮
│ # │  buf  │ 0.000 │ 0.001 │ 0.010 │ 0.020  │ 0.050  │ 0.100  │ 0.200  │ 0.300  │
├───┼───────┼───────┼───────┼───────┼────────┼────────┼────────┼────────┼────────┤
│ 0 │   128 │ 0%    │ 0%    │ 0%    │ 0%     │ 0%     │ 0%     │ 0%     │ 0%     │
│ 1 │   256 │ 0%    │ 0%    │ 0%    │ 0%     │ 0%     │ 0%     │ 0%     │ 0%     │
│ 2 │   512 │ 0%    │ 0%    │ 0%    │ 0%     │ 0%     │ 0%     │ 0%     │ 0%     │
│ 3 │  1024 │ 0%    │ 0%    │ 0%    │ 0%     │ 0%     │ 0%     │ 0%     │ 0%     │
│ 4 │  2048 │ 0%    │ 0%    │ 0%    │ 0%     │ 0%     │ -0.3%  │ 0%     │ +1%    │
│ 5 │  4096 │ 0%    │ 0%    │ 0%    │ -0.1%  │ +1.3%  │ -3.4%  │ +9.5%  │ +14.8% │
│ 6 │  8192 │ 0%    │ -0.6% │ -2.8% │ +0.2%  │ -2.9%  │ +11.3% │ +27.1% │ +5.5%  │
│ 7 │ 16384 │ 0%    │ -0.1% │ -8%   │ -16.9% │ -27.1% │ -35.3% │ +0.1%  │ -10.1% │
│ 8 │ 32768 │ 0%    │ +0.1% │ -16%  │ -33.7% │ -40.9% │ -49%   │ -4.4%  │ -41.4% │
╰───┴───────┴───────┴───────┴───────┴────────┴────────┴────────┴────────┴────────╯

Multiflow:

╭───┬───────┬──────────┬──────────┬──────────┬──────────┬────────╮
│ # │ flows │ agg_thru │ min_thru │ max_thru │ fairness │ drops  │
├───┼───────┼──────────┼──────────┼──────────┼──────────┼────────┤
│ 0 │     1 │ 0%       │ 0%       │ 0%       │ 0%       │ 0%     │
│ 1 │     2 │ 0%       │ 0%       │ 0%       │ 0%       │ 0%     │
│ 2 │     4 │ 0%       │ 0%       │ 0%       │ 0%       │ 0%     │
│ 3 │    16 │ +8.7%    │ +20.9%   │ +9.8%    │ +5.9%    │ -74.8% │
│ 4 │    32 │ +7.5%    │ +61.5%   │ -15.9%   │ +11.4%   │ -67.9% │
│ 5 │    64 │ +7.1%    │ +41.8%   │ -24.5%   │ +10.9%   │ -58.7% │
╰───┴───────┴──────────┴──────────┴──────────┴──────────┴────────╯

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.52%. Comparing base (e347a1e) to head (235c8c7).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1155      +/-   ##
==========================================
+ Coverage   81.48%   81.52%   +0.03%     
==========================================
  Files          81       81              
  Lines       25007    25045      +38     
==========================================
+ Hits        20378    20417      +39     
+ Misses       4629     4628       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Dirbaio]

it's possible self.cwnd_remaining() is 0, which would cause effective_mss to be 0. I think fast retransmit needs to be exempted from cc?