Fix acronym snake case in generated C names

[tolgahan-arikan]

What changed

• normalized acronym boundaries before snake-casing generated C symbols
• updated type, enum, service, method, and field naming paths to keep names like EOAWalletData, OIDCAuth, and ExchangeID split correctly
• added a regression test covering generated type, field, and method names for acronym-heavy schema identifiers

Why

The generator was feeding raw schema names directly into snakeCase. For identifiers with an uppercase acronym followed by a normal PascalCase word, that produced broken splits such as EOAWalletData -> eoaw_allet_data.

Impact

This fixes the generated C client surface so symbol names and struct member names are consistent and readable. It does not change wire format, request paths, JSON field names, enum wire values, or transport behavior.

Validation

• go test ./...