Keep tracking the file offset in memory, don't perform expensive Seek calls on every Read|WriteAsync

[adamsitnik]

This PR is blocked because it contains #48813 which needs to be reviewed and merged first.

Until then if someone is interested in looking at the changes you can see the three following commits:
d06f175
91423fa
a7ca4cb

But you should most probably just ignore it until then.

Depending on the file and buffer size, ReadAsync is 10-30% faster:

Method	Toolchain	fileSize	userBufferSize	options	Mean	Ratio	Allocated
ReadAsync	\after\CoreRun.exe	1024	4096	None	64.79 us	1.00	792 B
ReadAsync	\before\CoreRun.exe	1024	4096	None	65.04 us	1.00	792 B
ReadAsync	\after\CoreRun.exe	1024	4096	Asynchronous	77.38 us	0.92	953 B
ReadAsync	\before\CoreRun.exe	1024	4096	Asynchronous	83.77 us	1.00	952 B
ReadAsync	\after\CoreRun.exe	1048576	512	None	1,613.98 us	0.99	86,723 B
ReadAsync	\before\CoreRun.exe	1048576	512	None	1,625.09 us	1.00	86,721 B
ReadAsync	\after\CoreRun.exe	1048576	512	Asynchronous	3,240.19 us	0.70	95,047 B
ReadAsync	\before\CoreRun.exe	1048576	512	Asynchronous	4,632.78 us	1.00	95,047 B
ReadAsync	\after\CoreRun.exe	1048576	4096	None	1,054.26 us	1.00	29,353 B
ReadAsync	\before\CoreRun.exe	1048576	4096	None	1,051.53 us	1.00	29,353 B
ReadAsync	\after\CoreRun.exe	1048576	4096	Asynchronous	3,267.70 us	0.71	80,514 B
ReadAsync	\before\CoreRun.exe	1048576	4096	Asynchronous	4,617.27 us	1.00	80,514 B
ReadAsync	\after\CoreRun.exe	104857600	4096	None	126,194.78 us	1.01	2,868,024 B
ReadAsync	\before\CoreRun.exe	104857600	4096	None	125,496.57 us	1.00	2,868,024 B
ReadAsync	\after\CoreRun.exe	104857600	4096	Asynchronous	360,386.97 us	0.77	7,988,704 B
ReadAsync	\before\CoreRun.exe	104857600	4096	Asynchronous	470,759.18 us	1.00	7,991,376 B

And depending on file and buffer size WriteAsync is 0 to 70% faster:

Method	Toolchain	fileSize	userBufferSize	options	Mean	Ratio	RatioSD	Allocated
WriteAsync	\after\CoreRun.exe	1024	4096	Asynchronous	188.7 us	1.01	0.02	304 B
WriteAsync	\before\CoreRun.exe	1024	4096	Asynchronous	186.6 us	1.00	0.00	304 B
WriteAsync	\after\CoreRun.exe	1048576	512	Asynchronous	8,384.7 us	0.53	0.01	86,833 B
WriteAsync	\before\CoreRun.exe	1048576	512	Asynchronous	15,923.3 us	1.00	0.00	86,858 B
WriteAsync	\after\CoreRun.exe	1048576	4096	Asynchronous	8,288.1 us	0.52	0.02	80,510 B
WriteAsync	\before\CoreRun.exe	1048576	4096	Asynchronous	16,087.7 us	1.00	0.00	80,538 B
WriteAsync_NoBuffering	\after\CoreRun.exe	1048576	16384	Asynchronous	26,965.8 us	0.94	0.01	20,416 B
WriteAsync_NoBuffering	\before\CoreRun.exe	1048576	16384	Asynchronous	28,785.2 us	1.00	0.00	20,418 B
WriteAsync	\6.0.0\CoreRun.exe	104857600	4096	Asynchronous	488,837.8 us	0.29	0.02	7,988,032 B
WriteAsync	\simple\CoreRun.exe	104857600	4096	Asynchronous	1,671,306.4 us	1.00	0.00	7,988,032 B
WriteAsync_NoBuffering	\6.0.0\CoreRun.exe	104857600	16384	Asynchronous	157,768.7 us	0.32	0.03	1,997,376 B
WriteAsync_NoBuffering	\simple\CoreRun.exe	104857600	16384	Asynchronous	498,606.8 us	1.00	0.00	1,997,432 B

There is no perf penalty for sync reads and writes:

Method	Toolchain	fileSize	userBufferSize	options	Mean	Ratio	Allocated
Read	\after\CoreRun.exe	1024	4096	None	49.66 us	1.00	256 B
Read	\before\CoreRun.exe	1024	4096	None	49.67 us	1.00	256 B
Read	\after\CoreRun.exe	1048576	512	None	714.12 us	1.00	4,376 B
Read	\before\CoreRun.exe	1048576	512	None	716.84 us	1.00	4,376 B
Read	\after\CoreRun.exe	1048576	4096	None	656.52 us	0.99	256 B
Read	\before\CoreRun.exe	1048576	4096	None	661.34 us	1.00	256 B
Read	\after\CoreRun.exe	104857600	4096	None	79,517.27 us	1.01	292 B
Read	\before\CoreRun.exe	104857600	4096	None	79,127.17 us	1.00	292 B

fixes #16354
fixes #25905

[adamsitnik]

I am closing the PR. Our plan is that @jozkee is going to pick it up and combine it with #49145 and send a new PR. In #49754 I've added few edge-case tests to help to ensure we are not breaking anything.