v0.2.0: Zero-overhead dispatch

What's New

Zero-overhead dispatch path

• ProcessBatchWith<Visitor>: Compile-time dispatch bypassing callback_table_ and shared_mutex entirely. 19-42% faster than ProcessBatch.
• StaticComponent<Derived, PayloadVariant>: CRTP zero-overhead component with HasHandler SFINAE trait detection. No virtual destructor, no shared_ptr, handler calls are inlineable.
• FixedFunction<Sig, Capacity>: SBO type-erased callable replacing std::function. Zero heap allocation, static_assert on overflow.

Breaking changes

• CallbackType changed from std::function<void(const EnvelopeType&)> to FixedFunction<void(const EnvelopeType&), 64>. Lambda captures must fit within 64 bytes.

Performance (MPSC default, x86-64)

Path	Latency	vs v0.1.0
ProcessBatchWith (Visitor)	129 ns/msg	New
ProcessBatch (Callback)	160 ns/msg	-
FULL_FEATURED enqueue	38 ns/msg	improved
BARE_METAL enqueue	30 ns/msg	improved
E2E Latency P99	525 ns	improved

Tests

• 171 test cases (was 70), all passing
• New test files: test_fixed_function, test_visitor_dispatch, test_static_component

Files

• 3 header files: mccc.hpp + component.hpp + static_component.hpp
• Header-only, zero external dependencies, pure C++17