Updated benchmark to use pooling

[MarkCiliaVincenti]

The way you are benchmarking currently doesn't cater for multi-threading or contention. I didn't want to change such tests; however, I did amend the benchmark for my library to use the pooling (which by default is disabled) and you will be able to notice the big difference in the memory allocations.

[usercode]

Yeah, I will add more benchmarks later.

After enabling the pooling, AsyncKeyedLock is slower than ImageSharpWeb. I think the main reason for pooling is to get a better performance and not to reduce memory. That's why all frameworks enable this feature per default.

[MarkCiliaVincenti]

Yeah, I will add more benchmarks later.

After enabling the pooling, AsyncKeyedLock is slower than ImageSharpWeb. I think the main reason for pooling is to get a better performance and not to reduce memory. That's why all frameworks enable this feature per default.

The main reason for the polling in AsyncKeyedLock is to reduce memory allocations. In the benchmarks I did, it also improved performance.