staking: Fix reward distribution when common pool is exhausted

[abukosek]

This PR fixes the scenario when the reward schedule hasn't completed yet, but the common pool has been exhausted.

[codecov]

Codecov Report

Merging #5319 (3d2aa62) into master (0961dcc) will decrease coverage by 0.33%.
The diff coverage is 74.36%.

❗ Current head 3d2aa62 differs from pull request most recent head f7246af. Consider uploading reports for the commit f7246af to get more accurate results

@@            Coverage Diff             @@
##           master    #5319      +/-   ##
==========================================
- Coverage   66.97%   66.64%   -0.33%     
==========================================
  Files         524      524              
  Lines       55370    55776     +406     
==========================================
+ Hits        37082    37173      +91     
- Misses      13771    14017     +246     
- Partials     4517     4586      +69     

Impacted Files	Coverage Δ
go/consensus/api/api.go	38.88% <ø> (ø)
go/consensus/cometbft/abci/state.go	76.94% <ø> (ø)
go/consensus/cometbft/common/common.go	53.33% <ø> (ø)
go/keymanager/api/policy_sgx.go	46.66% <ø> (+13.33%)	⬆️
go/registry/api/api.go	55.93% <ø> (+0.15%)	⬆️
go/runtime/host/mock/mock.go	87.40% <0.00%> (-1.40%)	⬇️
go/runtime/host/protocol/types.go	54.54% <ø> (ø)
go/runtime/host/sgx/sgx.go	68.75% <0.00%> (-0.62%)	⬇️
go/worker/keymanager/api/api.go	0.00% <ø> (-32.44%)	⬇️
go/worker/keymanager/metrics.go	100.00% <ø> (ø)
... and 22 more

... and 71 files with indirect coverage changes

[peternose]

If this condition is true (q > commonPool), should we lower the reward and give as much as we can (q = commmonPool.Clone())?

// If an error occurs, the pool and affected accounts are left in an invalid state.
// This may fail due to the common pool running out of stake.

The method's comment is not correct anymore, as the pool will never run out of stake.

This method is not fair. For example:

• If I have an unlimited escrow balance, my rewards will always be too big and therefore skipped while others will receive their awards until they drain the pool.
• Assuming equal balances and too small pool to cover all rewards, the first in the list will get more rewards than the last one.

However, I'm not sure if this is a problem. If we want fairness, we should distribute the reminder of the pool proportional to stakes. But that is probably too much.

[abukosek]

Yeah, we should discuss what to do about this and implement it in a separate PR. This PR just fixes the panic that would have happened otherwise :)