Issue841 remove file stream construction

[LoopedBard3]

Initial testing for #841. Focused specifically on the ReadByte case to figure out the pattern we want to use before trying to apply it to all of the tests.

[LoopedBard3]

ReadByte Default Benchmark Statistics Before & After:
Before:

Method	fileSize	options	Mean	Error	StdDev	Median	Min	Max	Allocated
ReadByte	1024	None	147.5 us	3.20 us	3.55 us	146.5 us	143.0 us	152.6 us	4 KB
ReadByte	1024	Asynchronous	153.8 us	2.64 us	2.71 us	153.8 us	149.9 us	159.0 us	4 KB

After:

Method	fileSize	options	Mean	Error	StdDev	Median	Min	Max	Allocated
ReadByte	1024	None	1.793 us	0.0347 us	0.0356 us	1.780 us	1.742 us	1.877 us	-
ReadByte	1024	Asynchronous	1.785 us	0.0220 us	0.0184 us	1.785 us	1.747 us	1.820 us	-

I also ran the same code but with two 16kibibyte sized files to ensure that something else wasn't causing the super low times (like accidentally only reading one byte):

Method	fileSize	options	Mean	Error	StdDev	Median	Min	Max	Allocated
ReadByte	1024	None	1.782 us	0.0216 us	0.0180 us	1.786 us	1.749 us	1.817 us	-
ReadByte	1024	Asynchronous	2.050 us	0.0479 us	0.0551 us	2.039 us	1.981 us	2.182 us	-
ReadByte	16384	None	36.135 us	0.4380 us	0.3657 us	36.118 us	35.392 us	36.692 us	-
ReadByte	16384	Asynchronous	55.781 us	0.8563 us	0.8009 us	55.801 us	54.565 us	57.342 us	864 B

[LoopedBard3]

@adamsitnik Going through some old issues in the perf repo and Bill mentioned your opinion on this change would be beneficial. What are your thoughts on using this general pattern for minimizing the creation of the FileStream during the tests? Thanks!

[billwert]

@adamsitnik ping