Alternative concurrent counter

[bitfaster]

Address feedback from .NET team.

Benchmark taken from https://github.com/dotnet/runtime/pull/91566

New counter vs striped is marginally slower in these tests, although likely noise:

BenchmarkDotNet v0.15.8, Windows 11 (10.0.26200.9106/25H2/2025Update/HudsonValley2)
Intel Xeon W-2133 CPU 3.60GHz, 1 CPU, 12 logical and 6 physical cores
.NET SDK 10.0.400
  [Host] : .NET 10.0.11 (10.0.11, 10.0.1126.37416), X64 RyuJIT x86-64-v4
  net9.0 : .NET 9.0.19 (9.0.19, 9.0.1926.36724), X64 RyuJIT x86-64-v4

Job=net9.0  Runtime=.NET 9.0

| Method                   | Mean      | Error     | StdDev    |
|------------------------- |----------:|----------:|----------:|
| CounterSerial            | 20.179 ms | 0.2076 ms | 0.1942 ms |
| CounterParallel          |  2.189 ms | 0.0335 ms | 0.0297 ms |
| Striped64CounterSerial   | 19.944 ms | 0.0514 ms | 0.0481 ms |
| Striped64CounterParallel |  2.195 ms | 0.0278 ms | 0.0247 ms |
| MetricsCounterSerial     | 65.934 ms | 0.8149 ms | 0.6805 ms |
| MetricsCounterParallel   | 52.809 ms | 0.4452 ms | 0.4164 ms |
| UpDownCounterSerial      | 67.020 ms | 1.0584 ms | 0.9383 ms |
| UpDownCounterParallel    | 42.196 ms | 0.8298 ms | 1.0191 ms |

[coveralls]

coverage: 99.228% (+0.02%) from 99.212% — users/alexpeck/counterplus into main

[bitfaster]

@copilot remove all the nullable ? annotations to fix the warnings in this file.

[Copilot]

Done in 199da0e - removed the ? annotations, which fixes the CS8632 warnings since nullable context isn't enabled in this project.