Reduce allocations of TextSerializer

[Stabzs]

This PR reduces the allocations of TextSerializer by converting double/long values to byte arrays via intermediary buffers instead of creating intermediary strings. It does this by taking advantage of UftFormatters from the System.Memory assembly. All unit tests continue to pass as expected.

SerializationBenchmarks showed a 50% allocation reduction for CollectAndSerialize, while it brought CollectAndSerializeOpenMetrics down 66% into alignment with CollectAndSerialize. Since the runs of both benchmarks are bimodal, runtimes are difficult to compare although in my tests, they were unchanged and/or showed no significant differences.

Existing:

BenchmarkDotNet=v0.13.2, OS=Windows 10 (10.0.19043.1826/21H1/May2021Update)
Intel Core i7-8565U CPU 1.80GHz (Whiskey Lake), 1 CPU, 8 logical and 4 physical cores
.NET SDK=7.0.201
  [Host]  : .NET 7.0.3 (7.0.323.6910), X64 RyuJIT AVX2
  LongRun : .NET 7.0.3 (7.0.323.6910), X64 RyuJIT AVX2

Job=LongRun  IterationCount=100  LaunchCount=3  
WarmupCount=15  

Method	Mean	Error	StdDev	Median	Gen0	Allocated
CollectAndSerialize	104.09 ms	3.845 ms	19.77 ms	102.44 ms	428.5714	2.07 MB
CollectAndSerializeOpenMetrics	99.72 ms	4.825 ms	24.37 ms	86.99 ms	500.0000	2.94 MB

Optimized:

BenchmarkDotNet=v0.13.2, OS=Windows 10 (10.0.19043.1826/21H1/May2021Update)
Intel Core i7-8565U CPU 1.80GHz (Whiskey Lake), 1 CPU, 8 logical and 4 physical cores
.NET SDK=7.0.201
  [Host]  : .NET 7.0.3 (7.0.323.6910), X64 RyuJIT AVX2
  LongRun : .NET 7.0.3 (7.0.323.6910), X64 RyuJIT AVX2

Job=LongRun  IterationCount=100  LaunchCount=3  
WarmupCount=15  

Method	Mean	Error	StdDev	Gen0	Allocated
CollectAndSerialize	95.12 ms	3.763 ms	18.97 ms	200.0000	1.01 MB
CollectAndSerializeOpenMetrics	94.29 ms	4.322 ms	22.22 ms	250.0000	1.01 MB

NOTE: there are potential opportunities to use UTF8 literals in other places to minimize allocations on startup, but that seems like a separate effort.

[Stabzs]

@sandersaares Is there anything you need me to take action on for this PR? As far as I can tell, the build issues are not related to my changes.

[slang25]

This looks really nice 👏 Hope we can see this in an upcoming release