CSHARP-5554: Improved ObjectId parsing and ToString() performance

[pieceofsummer]

While writing an ObjectId-like class for another project (where I needed a time-based identifier, but didn't want to bring dependency on a whole MongoDB.Bson package just for ObjectId), I was curious about performance of my implementation compared to the reference one.

My test was like serializing a million of identifiers (as strings), and then parsing them back. Here are my results:

	ObjectId.ToString()	ObjectId.Parse()
MongoDB.Bson (v2.3.0)	3475ms	897ms
My (humble) implementation	87ms	81ms

So this probably looks like a valid reason to make a pull request and improve things a little :)

I'm not sure if it affects other cases where ObjectId is serialized, like performance of database responses' deserialization. But shouldn't hurt at least.

[damieng]

Please ignore the previous comments/closure, was meant for a different PR.

[ArielTipalti]

Suggested change

	return obj is ObjectId && Equals((ObjectId)obj);
	return obj is ObjectId objId && Equals(objId);

[ArielTipalti]

Suggested change

	var buffer = new char[24];
	public override string ToString()
	{
	Span<char> buffer = stackalloc char[24];
	int a = _a;
	int b = _b;
	int c = _c;
	for (int i = 7; i >= 0; i--)
	{
	var r = (a & 0x0F);
	buffer[i] = (char)(r < 10 ? (r + '0') : (r - 10 + 'a'));
	a >>= 4;
	r = (b & 0x0F);
	buffer[i + 8] = (char)(r < 10 ? (r + '0') : (r - 10 + 'a'));
	b >>= 4;
	r = (c & 0x0F);
	buffer[i + 16] = (char)(r < 10 ? (r + '0') : (r - 10 + 'a'));
	c >>= 4;
	}
	return new string(buffer);
	}