Fastpath

[alanmcgovern]

When a ReusableTask continues from another ReusableTask we can
pass the StateMachineCache object directly as the continuation
to avoid creating an Action which points to the StateMachineCache.

This optimisation works for ReusableTask -> ReusableTask as well as ReusableTask<T> -> ReusableTask and ReusableTask -> ReusableTask<T>.

Additionally, this makes a bunch of Cache objects a little bit smaller
as they don't have to hold an Action object.

BenchmarkDotNet=v0.13.1, OS=Windows 10.0.22631
AMD Ryzen 5 5600U with Radeon Graphics, 1 CPU, 12 logical and 6 physical cores
.NET SDK=8.0.300
[Host] : .NET 6.0.30 (6.0.3024.21525), X64 RyuJIT
DefaultJob : .NET 6.0.30 (6.0.3024.21525), X64 RyuJIT

Before

Method	Concurrency	Depth	Iterations	Mean	Error	StdDev	Median	Gen 0	Gen 1	Allocated
ReusableTask	1	500	1000	64.40 ms	4.009 ms	11.757 ms	61.14 ms	-	-	32 KB
ReusableTaskInt	1	500	1000	40.76 ms	3.133 ms	8.733 ms	37.86 ms	-	-	32 KB
ValueTaskInt	1	500	1000	65.22 ms	1.281 ms	2.140 ms	65.50 ms	6300.0000	100.0000	51,452 KB
ReusableTask	4	500	1000	91.19 ms	3.628 ms	10.698 ms	92.22 ms	-	-	32 KB
ReusableTaskInt	4	500	1000	57.23 ms	2.273 ms	6.630 ms	57.10 ms	-	-	127 KB
ValueTaskInt	4	500	1000	154.20 ms	3.067 ms	5.835 ms	154.74 ms	22250.0000	1000.0000	182,428 KB

After

Method	Concurrency	Depth	Iterations	Mean	Error	StdDev	Median	Gen 0	Gen 1	Allocated
ReusableTask	1	500	1000	41.91 ms	1.705 ms	4.667 ms	41.04 ms	-	-	32 KB
ReusableTaskInt	1	500	1000	34.96 ms	5.059 ms	14.917 ms	33.78 ms	-	-	32 KB
ValueTaskInt	1	500	1000	61.91 ms	3.413 ms	9.738 ms	56.92 ms	6000.0000	-	49,869 KB
ReusableTask	4	500	1000	55.24 ms	2.077 ms	6.124 ms	54.68 ms	-	-	32 KB
ReusableTaskInt	4	500	1000	51.82 ms	2.425 ms	7.111 ms	51.74 ms	-	-	127 KB
ValueTaskInt	4	500	1000	157.49 ms	3.112 ms	8.199 ms	157.46 ms	22500.0000	1000.0000	182,908 KB