Fix BitArray.CopyTo to byte[] regression - update offset for extra bits.

[ahsonkhan]

Fixes dotnet/core#3194 and https://github.com/dotnet/corefx/issues/39929

We need to port this to release/3.0

cc @stephentoub, @jkotas, @safern, @MichaelL79, @carlossanlop, @Clockwork-Muse, @ericstj, @danielValdezR

[ahsonkhan]

Before the change from #33367, this used to be as follows:

corefx/src/System.Collections/src/System/Collections/BitArray.cs

Lines 545 to 554 in 8716f1a

	// copy all the perfectly-aligned bytes
	for (int i = 0; i < arrayLength; i++)
	b[index + i] = (byte)((m_array[i / 4] >> ((i % 4) * 8)) & 0x000000FF); // Shift to bring the required byte to LSB, then mask
	if (extraBits > 0)
	{
	// mask the final byte
	int i = arrayLength;
	b[index + i] = (byte)((m_array[i / 4] >> ((i % 4) * 8)) & ((1 << extraBits) - 1));
	}

Before, we were writing one byte at a time, and then if the bit array contained extra bits to be copied over, we used the next index to write those bits. Now, we are writing 4 bytes at a time, and if we have 1-3 bytes left to write, we have the switch on remainder below. Beyond that, any extra bits left to copy (i.e. < 1 byte) should go at offset remainder.

I believe using remainder as the offset to write the last bits to is correct (and the test pass). Please review/verify.

[EgorBo]

By the way

(byte)((m_array[quotient] >> (remainder * 8)) & ((1 << (int)extraBits) - 1));

can be optimized in JIT: https://godbolt.org/z/XQ9Mpe (or at least with Bmi2.ZeroHighBits)
current codegen: sharplab.io

[ahsonkhan]

cc @AndyAyersMS

[ahsonkhan]

/azp run corefx-ci

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[safern]

Are these comments necessary?

[safern]

I think the test name speaks for itself 😄

[safern]

LGTM, thanks @ahsonkhan