Measures the full C++ -> Nim -> C++ foreign round trip across setups:
scalar family (2 x int64 + float64), payload family (seq[byte], size
sweep incl. large payloads), and event delivery — through sync, async
(bounded in-flight window) and event lanes, over a thread-count sweep.
All handlers compute the same O(1) parity predicate, so setups differ
purely in transport cost. Every reply is verified; a bad reply or lost
event exits non-zero.
`nimble perf_cpp_e2e` regenerates the cpp bindings, builds libperfbench
with -d:danger over the NIM_FFI_MM matrix (orc + refc) — deliberately
bypassing the debug-orc nim_ffi_lib.cmake template build — and runs the
Release driver. NIM_FFI_PERF_* env knobs control threads, volume,
iterations and payload sizes; each table row also emits a csv line for
diff-friendly capture.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* protect against mem leak in case of failures sending requests to ffi thread
* better cleanup if failures in createFFIContext
* avoid dangling cstring in handleRes under ARC/ORC
* better resource cleanup in destroyFFIContext
* invoke onNotResponding if failure in destroyFFIContext
* correct seq copy in alloc
* make sure the lock is init before cleanUpResources
* better possible exception handling in processReq
* guard allocSharedSeq if given seq is empty
* enhance error handling in ffi_context
* add new tests and some corrections
