Volatile read LFU drain status

[bitfaster]

Volatile read before try enter lock. This is slightly faster in the benchmark.

[coveralls]

coverage: 99.157% (+0.1%) from 99.057%
when pulling 57ecc8e on users/alexpeck/volatile
into fcfd183 on main.

[bitfaster]

Method	Runtime	Mean	Error	StdDev	Ratio	Code Size	Allocated
ConcurrentDictionary	.NET 6.0	7.641 ns	0.0996 ns	0.0883 ns	1.00	1,523 B	-
FastConcurrentLru	.NET 6.0	9.401 ns	0.0196 ns	0.0164 ns	1.23	7,047 B	-
ConcurrentLru	.NET 6.0	14.806 ns	0.0686 ns	0.0642 ns	1.94	7,294 B	-
AtomicFastLru	.NET 6.0	20.827 ns	0.1498 ns	0.1251 ns	2.72	NA	-
FastConcurrentTLru	.NET 6.0	10.787 ns	0.0445 ns	0.0371 ns	1.41	7,190 B	-
FastConcLruAfterAccess	.NET 6.0	11.401 ns	0.0224 ns	0.0187 ns	1.49	7,596 B	-
FastConcLruAfter	.NET 6.0	13.062 ns	0.0392 ns	0.0348 ns	1.71	6,429 B	-
ConcurrentTLru	.NET 6.0	16.304 ns	0.0908 ns	0.0805 ns	2.13	7,705 B	-
ConcurrentLfu	.NET 6.0	23.732 ns	0.3340 ns	0.3124 ns	3.11	NA	-
ClassicLru	.NET 6.0	43.489 ns	0.1085 ns	0.0906 ns	5.69	NA	-
RuntimeMemoryCacheGet	.NET 6.0	123.112 ns	2.4499 ns	3.6670 ns	16.40	49 B	32 B
ExtensionsMemoryCacheGet	.NET 6.0	57.915 ns	1.1798 ns	3.0665 ns	7.00	78 B	24 B
ConcurrentDictionary	.NET Framework 4.8	13.464 ns	0.0610 ns	0.0541 ns	1.00	4,127 B	-
FastConcurrentLru	.NET Framework 4.8	14.947 ns	0.1641 ns	0.1535 ns	1.11	27,372 B	-
ConcurrentLru	.NET Framework 4.8	19.250 ns	0.1383 ns	0.1294 ns	1.43	27,676 B	-
AtomicFastLru	.NET Framework 4.8	32.326 ns	0.3353 ns	0.2972 ns	2.40	358 B	-
FastConcurrentTLru	.NET Framework 4.8	46.813 ns	0.7111 ns	0.6651 ns	3.48	27,664 B	-
FastConcLruAfterAccess	.NET Framework 4.8	45.892 ns	0.3912 ns	0.3660 ns	3.41	358 B	-
FastConcLruAfter	.NET Framework 4.8	47.184 ns	0.2052 ns	0.1819 ns	3.50	358 B	-
ConcurrentTLru	.NET Framework 4.8	48.380 ns	0.1636 ns	0.1450 ns	3.59	28,016 B	-
ConcurrentLfu	.NET Framework 4.8	40.882 ns	0.4661 ns	0.4132 ns	3.04	NA	-
ClassicLru	.NET Framework 4.8	50.470 ns	0.1725 ns	0.1529 ns	3.75	NA	-
RuntimeMemoryCacheGet	.NET Framework 4.8	286.450 ns	0.6913 ns	0.5772 ns	21.29	33 B	32 B