fix: daily blockchain tests

[DavePearce]

This fixes the issues running the daily blockchain tests, by doing two things:

1. Updating to the latest version of go-corset, which includes some minor optimisations that help reduce cost.
2. Adding a batching parameter for go-corset (-b 1024) which sets the batch size to max 1024. This throttles concurrency to some extent, and reduces memory pressure.

Overall, these changes appear to allow the daily blockchain tests to pass, and hopefully they will consistently pass going forward.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[OlivierBBB]

Approved. Where does the value 1024 come from and do you know how many batches the previously unthrottled implementation would use ? Also are these batches batches of columns that corset computes or something else ?

[DavePearce]

The 1024 figure is just one that seems to work well. I played around with a few. Without a batch size specified it tries to compute as many columns in one go as it can. Given that we have max ~2300
Columns, of which some are input columns and don't need to be computed ... i guess the largest batch probably isn't much larger than 1024. But the difference is obviously enough to tip it back into the green 🤔

[DavePearce]

Actually forget that. Its not columns causing the problem but constraints I guess. Which means it might be trying to do all 9000 in one go!! (Since constraints don't have dependencies between themselves)