Enable SkipLocalsInit for .NET6

[bitfaster]

[SkipLocalsInit] can be used to improve performance - https://github.com/dotnet/roslyn/blob/master/docs/features/skiplocalsinit.md

Runtime applies this attribute on all their assemblies for 5.0 - https://github.com/dotnet/runtime/search?q=SkipLocalsInit

[coveralls]

coverage: 98.58% (+0.04%) from 98.537% when pulling 92530ca on users/alexpeck/skipinit into 5c2a527 on main.

[bitfaster]

Summary:

• Code size is smaller
• ConcurrentDictionary was inexplicably fast for the SkipLocalsInit run below, hence ratio makes it look worse but Mean is better. This was the most comparable result I had for LFU.
• Latency change was not that repeatable but does seem generally faster when running the benchmarks over and over.
• Everything is a touch faster, biggest difference is for TLru and Atomic (likely they have more things on stack that are not zero'd).
• Hard to get stable numbers for LFU (some runs have higher deviation). Does also seem to be a touch faster.

Before

Method	Mean	Error	StdDev	Ratio	Code Size	Allocated
ConcurrentDictionary	7.487 ns	0.0159 ns	0.0141 ns	1.00	1,523 B	-
FastConcurrentLru	8.727 ns	0.0384 ns	0.0359 ns	1.16	7,312 B	-
ConcurrentLru	14.836 ns	0.0186 ns	0.0156 ns	1.98	7,559 B	-
AtomicFastLru	20.160 ns	0.0394 ns	0.0369 ns	2.69	NA	-
FastConcurrentTLru	11.006 ns	0.0459 ns	0.0383 ns	1.47	7,725 B	-
ConcurrentTLru	16.122 ns	0.0143 ns	0.0127 ns	2.15	7,989 B	-
ConcurrentLfu	27.731 ns	0.9711 ns	2.8632 ns	4.11	NA	-
ClassicLru	43.325 ns	0.0439 ns	0.0411 ns	5.79	NA	-

After (with SkipLocalsInit)

Method	Mean	Error	StdDev	Ratio	Code Size	Allocated
ConcurrentDictionary	7.089 ns	0.0445 ns	0.0416 ns	1.00	1,523 B	-
FastConcurrentLru	8.109 ns	0.0132 ns	0.0111 ns	1.14	7,235 B	-
ConcurrentLru	14.598 ns	0.0154 ns	0.0136 ns	2.06	7,482 B	-
AtomicFastLru	19.677 ns	0.0209 ns	0.0186 ns	2.77	NA	-
FastConcurrentTLru	11.362 ns	0.0421 ns	0.0373 ns	1.60	6,394 B	-
ConcurrentTLru	16.332 ns	0.0933 ns	0.0873 ns	2.30	7,912 B	-
ConcurrentLfu	26.238 ns	0.5544 ns	0.9997 ns	3.72	NA	-
ClassicLru	43.118 ns	0.0938 ns	0.0877 ns	6.08	NA	-