Use current BFP rounding mode for s390x integer-to-double conversion

[abhishekabhi73cc-design]

The previous implementation explicitly encoded round-to-nearest, ties-to-even (M3 = 4) for the s390x integer-to-double conversion.

This change switches to M3 = 0, allowing the instruction to use the current BFP rounding mode. Since the CLI initializes the floating-point environment to IEEE-754 round-to-nearest and managed code cannot modify it, this preserves the required semantics while relying on the runtime-configured floating-point control state.

Verified by the existing BigInteger conversion regression tests.

[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.

[tannergooding]

Why make this change? It seems like its strictly worse and risks introducing breakage for scenarios where the floating-point control word is modified.

If there is a form that allows strictly specifying the dedicated rounding mode per instruction, that seems like pure goodness (and on other architectures can actually be a minor perf improvement).