block: let a header own its AuxPoW proof

[xanimo]

Prerequisite for the AuxPoW follow-up sequence (parse/check split, then an
AuxPoW serializer). Independent of the BIP157/BIP152 stacks — this branches
straight off 0.1.5-dev.

dogecoin_block_header carried an auxpow member holding only the validation
hook — check, ctx, is — never the proof. The proof was built into a local
dogecoin_auxpow_block inside dogecoin_block_header_deserialize and freed at
cleanup:, so by the time a caller had its header, the parent coinbase, merkle
branches and parent header were gone.

dogecoin_block_header_copy made it worse by looking complete — it copied the
three hook fields, so a copied merge-mined header was silently missing its proof
with nothing to indicate it.

Anything needing the proof after the parse therefore had to re-parse the wire
bytes or retain them separately. That is why BIP152 keeps a header_raw span
on dogecoin_compact_block — not because a compact block wants raw bytes, but
because the parsed header could not answer for itself.

What this adds

dogecoin_auxpow_payload, owned by the header. The deserializer moves the
parsed fields onto it and nulls them on the scratch block, so ownership
transfers rather than duplicating; _free releases it; _copy deep-copies it.

The payload deliberately has no back-pointer to its header, unlike
dogecoin_auxpow_block, which owns both its header and its parent_header
and frees them — a header holding one of those would own the thing that owns it.
The payload owns only parent_header, a plain 80-byte header whose own payload
is NULL, so ownership terminates.

auxpow.check / auxpow.ctx are untouched. That is a validation hook whose
context the caller supplies at call time — validation.c:111 passes the block
directly — not a reference to this data.

Test

Extends the height-371338 vector: the proof survives the parse, a copy owns an
independent one, and freeing the source leaves the copy intact — which it can
only do if nothing is shared.

Checked for bite, not just for passing: disabling only the deep copy in
_copy and leaving retention intact fails at line 351. A test that passed
either way would have proved nothing.

WITH_NET=ON with -DBUILD_SHARED_LIBS=1: 78/78. WITH_NET=OFF: 72/72.

What it unblocks

• parse/check split — deferring check_auxpow needs the proof to still be
  around after parsing, which it now is
• AuxPoW serializer — needs a header that can answer for its own proof; this
  is what makes one writable, and the encode side of
  dogecoin_compact_block_serialize needs it before libdogecoin can originate a
  cmpctblock for a block it assembled rather than received
• header_raw becomes an optimisation rather than a necessity

[edtubbs]

ACK, compact blocks give a reason to keep auxpow proofs around after validation, which is good for other uses too.