Avoid inlining Random.InternalSample

[AndyAyersMS]

This method contains an inherently unpredictable branch that benefits from if conversion. If inlined into a caller with a loop we lose the if conversion and performance suffers, and there does not appear to be other inlining benefit.

We can reconsider if/when we've freed up if-conversion from its current "not in loop" constraint.

Fixes #117787
Fixes #130359

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).
13 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-runtime
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This PR adds an explicit JIT hint to keep Random’s compatibility PRNG InternalSample as a standalone method, aiming to avoid unfavorable codegen when the method gets inlined into hot loops.

Changes:

• Add [MethodImpl(MethodImplOptions.NoInlining)] to CompatPrng.InternalSample().
• Add an explanatory comment documenting why the method should not be inlined.

[AndyAyersMS]

Local Windows x64 results:

Benchmark	Baseline	Changed
#117787, TieredPGO=1	821.4 µs	279.6 µs
#117787, TieredPGO=0	529.2 µs	541.9 µs (same)

All 40 dotnet/performance Perf_Random cases showed no regressions; NextDouble improved 40% and NextSingle 49%.

Also improves performance of #130359

Iterations	Baseline	NoInlining
1–10	53.1 ms	52.5 ms
11–20	48.3 ms	49.4 ms
21–30	48.7 ms	49.3 ms
31–40	93.0 ms	46.8 ms
41–50	123.3 ms	45.6 ms
51–60	122.8 ms	45.7 ms
61–70	121.3 ms	46.2 ms
71–80	120.1 ms	46.8 ms
81–90	120.2 ms	46.8 ms
91–100	119.7 ms	46.6 ms

@EgorBo PTAL
fyi @dotnet/jit-contrib