Verifies that multi-byte UTF-8 characters (Korean, emoji, accented
Latin, Japanese) are correctly preserved through the STDIO inbound
processing pipeline.

Use natural script names (한글, 漢字) and a common accented
word (café) for UTF-8 encoding test data.

The test passes regardless of whether UTF-8 is explicitly specified,
because the test environment default charset is already UTF-8.

Spawns a child JVM with -Dfile.encoding=ISO-8859-1 to simulate a
non-UTF-8 default charset environment. The test sends a JSON-RPC
message containing multi-byte characters (Korean, Chinese, accented
Latin, emoji) through StdioServerTransportProvider and verifies they
are preserved. This test fails without the explicit UTF-8 charset
fix and passes with it.

…tocol#295

The issue only reports encoding inconsistency in
StdioServerTransportProvider. Revert StdioClientTransport to its
original state to keep the PR scoped to the reported problem.