reduce allocations

[shawjef3]

This reduces allocations by using a Stack instead of a List for ChannelExecutor's doneStack. I measured the difference by running ZStream.unit.forever.runDrain and profiling the same way I did at #7365.

benchmarks

before

[info] Benchmark                                   (chunkCount)  (chunkSize)   Mode  Cnt  Score    Error  Units
[info] StreamBenchmarks.zioChunkChunkFilterMapSum         10000         5000  thrpt   15  3.299 ±  0.026  ops/s
[info] StreamBenchmarks.zioChunkFilterMapSum              10000         5000  thrpt   15  2.466 ±  0.055  ops/s
[info] StreamBenchmarks.zioGroupAdjecentBy                10000         5000  thrpt   15  1.083 ±  0.003  ops/s
[info] StreamBenchmarks.zioGroupByKey                     10000         5000  thrpt   15  2.533 ±  0.021  ops/s
[info] StreamBenchmarks.zioGroupWithin                    10000         5000  thrpt   15  0.033 ±  0.001  ops/s
[info] StreamBenchmarks.zioMapAccum                       10000         5000  thrpt   15  2.227 ±  0.020  ops/s
[info] StreamBenchmarks.zioSliding                        10000         5000  thrpt   15  0.117 ±  0.001  ops/s
[info] StreamBenchmarks.zioTakeWhile                      10000         5000  thrpt   15  3.212 ±  0.013  ops/s

after

[info] Benchmark                                   (chunkCount)  (chunkSize)   Mode  Cnt  Score   Error  Units
[info] StreamBenchmarks.zioChunkChunkFilterMapSum         10000         5000  thrpt   15  3.323 ± 0.042  ops/s
[info] StreamBenchmarks.zioChunkFilterMapSum              10000         5000  thrpt   15  2.512 ± 0.012  ops/s
[info] StreamBenchmarks.zioGroupAdjecentBy                10000         5000  thrpt   15  1.073 ± 0.003  ops/s
[info] StreamBenchmarks.zioGroupByKey                     10000         5000  thrpt   15  2.453 ± 0.187  ops/s
[info] StreamBenchmarks.zioGroupWithin                    10000         5000  thrpt   15  0.032 ± 0.001  ops/s
[info] StreamBenchmarks.zioMapAccum                       10000         5000  thrpt   15  2.194 ± 0.012  ops/s
[info] StreamBenchmarks.zioSliding                        10000         5000  thrpt   15  0.108 ± 0.001  ops/s
[info] StreamBenchmarks.zioTakeWhile                      10000         5000  thrpt   15  3.748 ± 0.025  ops/s

allocations

before

after

Notice that the $colon$colon allocations disappear, and nothing takes their place.