Adding try_state hook for Treasury pallet

[wentelteefje]

This PR adds a try_state hook for the Treasury pallet.
Part of #239.
The invariants checked are:

1. ProposalCount must be $\geq$ number of elements in Proposals. The ProposalCount storage value is only increased, but never decreased, whereas individual proposals can be removed from the Proposals storage map.
2. T::ProposalBondMinimum * Number of proposals with non-zero bond $\leq \sum_{p \in \textrm{Proposals}} \textrm{p.bond}$.

Since the introduction of the dispatchable spend, a bond is no longer required (bond is set to zero in these cases). However, the old way of proposing using a bond is still available and we can filter for these by looking at proposals with $\textrm{p.bond} \neq 0 $.

3. (Optional) T::ProposalBondMaximum * Number of proposals with non-zero bond $\geq \sum_{p \in \textrm{Proposals}} \textrm{p.bond}$.

Analogous to 2., but only relevant when T::ProposalBondMaximum is not None.

4. Each ProposalIndex contained in Approvals should exist in Proposals. Note, that this
   automatically implies Approvals.count() <= Proposals.count().
5. SpendCount must be $\geq$ number of elements in Spends. Follows the same reasoning as invariant 1.
6. For each spend entry contained in Spends we should have spend.expire_at > spend.valid_from.

[ggwpez]

Looks very clean, thanks!
Should probably try this before merging on Polkadot and Kusama states, so we know that it all works.

[ggwpez]

bot fmt
It is cargo +nightly-2023-05-16 fmt if you want to run it locally.

[wentelteefje]

cargo fmt seems to break markdown lists in comments. I guess I should fix these manually before merging too.

[ggwpez]

cargo fmt seems to break markdown lists in comments. I guess I should fix these manually before merging too.

Yea... lines should wrap at exactly 100 chars - also comments.

[wentelteefje]

bot fmt

[wentelteefje]

Looks very clean, thanks! Should probably try this before merging on Polkadot and Kusama states, so we know that it all works.

It works on rococo state (ProposalCount = 15), but because of the migration of Polkadot/Kusama runtimes into the fellowship's runtime repository I wasn't able to try it out on them.

[liamaharon]

looking good, thanks!

[wentelteefje]

I have added tests for all try_state scenarios. During this I also noticed that there's no longer a way to check invariants 2 & 3 from above (my initial approach had a mistake). That's because faulty entries with a zero bond cannot be distinguished from valid entries made using the spend_local dispatchable. So it was good to think about that some more during test writing :) Also tested successfully against rococo state once more just in case.

As for the tests: I really don't like catching errors based on a string value. An enum would be much cleaner, but I guess there's no other way for now?

[liamaharon]

Thanks, tests look great 👌

Other must be a String afaik. You might be able to avoid duplication be declaring them as consts, but not a big deal for the purposes of try-state.