kernel: expose contextual block validation (`TestBlockValidity()`) #35726

The kernel library currently exposes two API for block validation:

• btck_block_check : context-free checks only, and
• btck_chainstate_manager_process_block : full validation that also connects the block, persists it to disk, updates indexes and fires notifications.

We could expose TestBlockValidity() using btck_chainstate_manager_test_block_validity that would fully validate a block in the context of current tip and that doesn't commit it. This could be beneficial for mining softwares and for testing purpose that would do contextual header/block checks and full script/UTXO validation as read only operation.

<!--e57a25ab6845829454e8d69fc972939a-->

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

<!--006a51241073e994b41acfe9ec718e94-->

Code Coverage & Benchmarks

For details see: https://corecheck.dev/bitcoin/bitcoin/pulls/35726.

<!--021abf342d371248e50ceaed478a90ca-->

Reviews

See the guideline and AI policy for information on the review process. A summary of reviews will appear here.

<!--5faf32d7da4f0f540f40219e4f7537a3-->

Won't hurt, but I am concept ~0 motivation is not satisfying.

This could be beneficial for mining softwares

I think mining software uses the mining interface, which exposes checkblock. How is this better than that? Even the exposed checkblock IPC interface is likely not going to be used because it's context-heavy and not ideal when you want to validate a lot of block template proposals really fast. There is a better idea #35671 that is bandwidth-efficient and faster.

for testing purpose that would do contextual header/block checks and full script/UTXO validation as read only operation.

Seems useful, but still weak motivation imo, I'd rather we expose #35187, technically u can simulate the testing purpose you describe here with #35187, and it has a very important usecase which is validating block without the context of the current tip.

I think mining software uses the mining interface, which exposes checkblock. How is this better than that? Even the exposed checkblock IPC interface is likely not going to be used because it's context-heavy and not ideal when you want to validate a lot of block template proposals really fast. There is a better idea #35671 that is bandwidth-efficient and faster.

I think that makes sense, will lean towards #35187. Thanks, closing this for now.