Winternitz Improvements in Generic Format

[just-erray]

This PR optimizes the previous winternitz algorithms from winternitz.rs (ListpickVerifier) and winternitz_compact.rs (BruteforceVerifier), adds small improvements and fixes. Furthermore it adds two new algorithms (BinarysearchVerifier and HybridVerifier) for signature verification, as well as two new conversion algorithms (TabledConverter and StraightforwardConverter) from digit form to bytes, for the usage of various values of log_d. The table below shows each algorithms script size including signing in bytes for different message lengths, with the best suitable conversion algorithm.

20 bytes	log_d=4	log_d=5	log_d=6	log_d=7	log_d=8
ListpickVerifier	4178	5133	6245	8801	11862
BruteforceVerifier	7654	11310	17507	28955	50978
BinarysearchVerifier	4706	5483	6125	7735	8964
HybridVerifier	4486	5343	6005	7631	8895

32 bytes	log_d=4	log_d=5	log_d=6	log_d=7	log_d=8
ListpickVerifier	6458	8879	10125	13713	18031
BruteforceVerifier	11830	18766	27389	44719	77558
BinarysearchVerifier	7274	8804	9423	11981	13621
HybridVerifier	6934	8580	9239	11821	13516

128 bytes	log_d=4	log_d=5	log_d=6	log_d=7	log_d=8
ListpickVerifier	24698	30972	36676	51312	67415
BruteforceVerifier	45238	67875	102345	168367	290198
BinarysearchVerifier	27818	33062	35976	45121	50909
HybridVerifier	26518	32226	35276	44517	50516

256 bytes	log_d=4	log_d=5	log_d=6	log_d=7	log_d=8
ListpickVerifier	Crashed	61407	72564	100852	133267
BruteforceVerifier	89957	134502	202411	331076	573719
BinarysearchVerifier	Crashed	65547	71180	88675	100633
HybridVerifier	Crashed	63891	69796	87487	99856

[sander2]

This PR optimizes the previous winternitz algorithms from winternitz.rs (ListpickVerifier) and winternitz_compact.rs (BruteforceVerifier), adds small improvements and fixes.

Out of curiosity, could you expand on the improvements made compared to winternitz.rs / ListpickVerifier? As far as I can tell the script size remains unchanged, right?

Anyway, with this PR, would the two older files be completely redundant? If so, I'd be in favor of removing the old files

[just-erray]

Out of curiosity, could you expand on the improvements made compared to winternitz.rs / ListpickVerifier? As far as I can tell the script size remains unchanged, right?

For log_d=4, Listpickverifier has the exact script size as the previous winternitz.rs. For larger log_d values, it can only save 1 or 2 opcodes per digit because it duplicates the number instead of pushing it twice. Other algorithms are useful when the message size is large where we would want to choose larger log_d values considering maximum stack depth.

Actually the binary search algorithm was sitting in our repository for some time now and I was considering merging the two algorithms for a more optimized one (making HybridVerifier, it uses binary search and listing-OP_PICK algorithms consequently) , but the real inspiration came when I saw your PR (it is now an optimized special case of the current Hybrid algorithm, where it uses only one step of binary search), so thank you very much for that.

Anyway, with this PR, would the two older files be completely redundant?

I believe so.

[lucidLuckylee]

Anyway, with this PR, would the two older files be completely redundant? If so, I'd be in favor of removing the old files

Me too. Could you do that before we merge?

[just-erray]

Me too. Could you do that before we merge?

I think, in that case, the old winternitz_hash.rs file stops working due to its dependencies on the winternitz.rs file. Should I edit winternitz_hash.rs to make it compatible with the generic version? Also doing these edits doesn't seem to cause any issues with the current chunker projects. Alpen's chunker uses a version of their own, and I believe this one doesn't have a winternitz implementation yet.

What should I name the new code file if I delete the two previous ones? Should it be just winternitz.rs or winternitz_generic.rs?"

[lucidLuckylee]

Should I edit winternitz_hash.rs to make it compatible with the generic version?

That would be great.

Should it be just winternitz.rs or winternitz_generic.rs?

Just winternitz.rs seems good to me.

[lucidLuckylee]

Thanks @just-erray! Great work!