Keep protocol primitives native to bitcoin-rs

[gosunuts]

crates/primitives will remain a native bitcoin-rs implementation rather than adopting rust-bitcoin canonical types.

The expected benefit from replacing Tx, Block, Header, OutPoint, hashes, sighash, encoding, network types, and related primitives with rust-bitcoin equivalents is not large enough to justify the migration cost, compatibility layer, and tighter dependency coupling.

Principles:

• Keep protocol/domain primitives implemented and owned by bitcoin-rs.
• Remove rust-bitcoin dependencies from crates/primitives where they exist solely to provide or mirror Bitcoin primitive types, encoding, sighash, network, hash, transaction, block, header, or outpoint functionality.
• Do not add wrapper or conversion layers solely for rust-bitcoin compatibility.
• Keep storage/UTXO-specific packed representations where useful.
• External libraries may still be used where they provide clear standalone value, but they should not define bitcoin-rs's core primitive vocabulary.

The target state is that crates/primitives is self-contained with respect to Bitcoin protocol primitives and does not depend on rust-bitcoin for those responsibilities.

RPC compatibility is a separate decision and is intentionally out of scope for this issue.

[metaphorics]

@cursor aggressively optimize and enhance it very breaking neat way; batched multiple stacked PRs publications; Best effort from end to end.

[cursor]

I published the third stacked PR: #597 (Give Tx and Header native field newtypes), based on #522, which sits on #483.

Protocol fields are now bitcoin-rs newtypes (Amount, Sequence, Script, Witness, LockTime, CompactTarget) instead of raw integers and byte vectors. JSON/RPC and packed UTXO layouts still use satoshis/u32 at those boundaries. The rest of the workspace is updated; rust-bitcoin stays only at RPC/wallet-facing and differential-oracle seams.

I’m waiting on CI for #597 and on any review comments.