Echo-10K Benchmark: Validation Checks

Every check is byte-exact. The profile is an echo, so what has to be established is that the bytes come back unchanged - not that a length or a count matches.

Check	What it establishes
POST /echo with a 1-byte random body	the smallest possible body round-trips
POST /echo with a 1 KB random body	ordinary case
POST /echo with a 10 KB random body	the benchmark's own size
POST /echo with a 100 KB random body	a body larger than the benchmark uses, spanning several TLS records and socket buffers
POST /echo with a chunked body	the body is read from the chunked framing, not from a Content-Length
POST /echo with an empty body	200 with an empty response, not a 411 or a hang
TLS posture probe on :8081	ALPN negotiates http/1.1

The bodies are random, and that is deliberate. A fixed body can be answered from a canned response without ever reading the request. The benchmark rotates eight distinct bodies for the same reason; these checks make answering-without-reading impossible rather than merely unlikely.

The chunked probe is the one the old Upload profile never had. Every one of its checks sent a body with an accurate Content-Length and compared the returned byte count, so a handler that echoed that header without reading a single body byte passed the entire suite - and would have outranked one that actually did the work.