Support increased number of nodes in coordinator tests

[piotr-roslaniec]

• Fixes Providers not sorted thrown in test_coordinator.py due to case-insensitive sorting of providers
• Based on Beefing up the Coordinator: Restrictions to initiator, ritual duration, fee model, automatic TX reimbursements, etc #86
• Need 2 reviews

[cygnusv]

• Fixes Providers not sorted thrown in test_coordinator.py due to case-insensitive sorting of providers

Good catch!

• Increases the number of test accounts to 67 (64 nodes + 3 technical accounts)

Not sure if increasing the number of nodes in the test is really necessary in the long-term. In this current shape, it's just going to make the overall test time to be longer. Our protocol, from the point of view of node operations to the Coordinator, is constant-time (constant-gas, if we're nit-picky) except for:

• Transcript size: This is what could really benefit from make it dependent on the cohort size, but we would need to have a clear formula for this (or at least, a good approximation)
• the function getParticipantFromProvider(), which I'm planning to make constant-time in a forthcoming PR

So, I'd propose to keep the test node count low (anything between 4 and 8 is fine) and make the transcript size depend on the node count.

[piotr-roslaniec]

I agree with your analysis. I reverted the number of accounts and added a dynamic transcript size. For now, it uses an approximated formula and in the future, we can replace the actual ferveo::Transcript creation if needed.

[manumonti]

LGTM! 🙌

Maybe we should edit the description of this PR since AFAIK we are not increasing the number of test accounts to 67 with these changes.

• Increases the number of test accounts to 67 (64 nodes + 3 technical accounts)