Convert GetCurrentMethod to QCALL

[AustinWise]

This has a similar structure to FailFast as converted in #98908.

Contributes to #95695

[jkotas]

I see about 6% perf regression in MethodBase.GetCurrentMethod with this change (Win x64).

using System.Reflection;
using System.Diagnostics;

for(;;)
{
    Stopwatch sw = new Stopwatch();
    sw.Start();
    for (int i = 0; i < 1000000; i++) MethodBase.GetCurrentMethod();
    Console.WriteLine(sw.ElapsedMilliseconds.ToString());
}

The implementation of MethodBase.GetCurrentMethod comes with a lot of unnecessary ceremony. Eliminating the ceremony should fix the regression - it can be just:

        public static MethodBase? GetCurrentMethod()
        {
            StackCrawlMark stackMark = StackCrawlMark.LookForMyCaller;
            RuntimeMethodHandleInternal methodHandle = InternalGetCurrentMethod(new StackCrawlMarkHandle(ref stackMark));
            return methodHandle.IsNullHandle() ? null : RuntimeType.GetMethodBase(null, methodHandle);
        }

[AustinWise]

I was able to reproduce the regression in a benchmark. Interestingly, it appears that something between .NET 9 Preview 2 and the merge base sped things up:

dotnet/runtime	Mean	Error	StdDev
.NET 9 Preview 2	698.8 ns	2.27 ns	2.12 ns
This PR	639.9 ns	3.62 ns	3.38 ns
Merge Base	571.2 ns	2.98 ns	2.79 ns

Anyways, I'll mark this as draft while I fix the regression.

[AustinWise]

@jkotas Thanks for the tip, this PR is now a performance improvement. Benchmark results of MethodBase.GetCurrentMethod:

code	Mean	Error	StdDev
this PR	480.5 ns	6.14 ns	5.13 ns
merge base	571.2 ns	2.98 ns	2.79 ns
.NET 9 preview 2	698.8 ns	2.27 ns	2.12 ns

[jkotas]

Thank you!