[KeyManager] End-to-end FFI error handling standardization

[NilanjanDaw]

This PR performs a comprehensive refactoring of the KeyManager component to utilize the standardized keymanager.Status Protobuf enum across all layers of the system. It replaces legacy negative i32 error codes and ad-hoc error enums with a unified, type-safe Status reporting mechanism that spans Rust, C, and Go.

Key Changes

• Rust FFI & Core Updates:
  • Refactored KeyCustodyCore in both key_protection_service and workload_service to return the Status enum directly.
  • Simplified FFI implementation by utilizing the ffi_call and ffi_call_i32 helpers introduced in Part 1.
  • Updated internal logic to map crate-specific errors to the standardized Status proto.
• C-Binding & Header Standardization:
  • Modified C function signatures in kps_key_custody_core.h and ws_key_custody_core.h to return Status instead of int32_t.
• Go CGO & Service Layer Integration:
  • Updated CGO wrappers to interpret the new Status return codes and convert them into idiomatic Go errors using theToStatus() helper.
  • HTTP Status Mapping: Implemented httpStatusFromError in the Go server to automatically map standardized FFI errors to appropriate HTTP status codes (e.g., ERROR_NOT_FOUND → 404 Not Found, ERROR_INVALID_ARGUMENT → 400 Bad Request).
  • Removed redundant error aliases in types.go to maintain a single source of truth.
• Test Alignment: Updated integration tests in Go and unit tests in Rust to verify the new error reporting behavior and
  ensure consistent error messages (e.g., "FFI error: ERROR_NOT_FOUND").

[pgonda]

This is gonna get old.

Should you add a .ok() or similar method for:

if !keymanager.Status(rc).ok() {
...
}

[pgonda]

This is fine. Once you update to gRPC service this gets much easier

[pgonda]

Approval but please have someone give the rust code a quick look