[VS Incentives]: Remove liquidity-based reward splitting from CL

[AlpinYukseloglu]

Background

To aid with smoothing out the migration process from Classic to Concentrated pools, we implemented a light feature that approximated splitting rewards between linked Classic and Concentrated pools based on liquidity. The goal was to treat the linked Classic pool as a full range Concentrated position for the purposes of incentives, and kick back the appropriate amount of incentives back to the Classic pool's internal gauge with a small discount to nudge people to migrate.

While this was a very useful and elegantly implemented feature, I propose we remove it in favor of volume splitting incentives for the following reasons:

1. Volume splitting incentives are better in essentially all cases
2. The existence of automatic liquidity-based splitting between linked pools will interfere with volume-splitting incentives
3. Early CL pools have demonstrated that volume flocks to the concentrated pool quickly, so we should expect the incentives to be weighted heavily towards the CL pool and migrating should remain a rational thing to do

The second point is particularly difficult to circumvent because the liquidity-based splitting is implemented at the CL pool level of abstraction out of necessity, as an epoch-based model at the gauge layer would have led to weird games (compared to spoofing volume, spoofing liquidity is very low-cost).

Suggested Design

Fortunately, the feature was quite well-abstracted, so the actual change (outside of fixing tests) should actually be quite small.

In fact, the entire feature should be able to be disabled by simply removing the following two lines from updateGivenPoolUptimeAccumulatorsToNow:

osmosis/x/concentrated-liquidity/incentives.go

Line 390 in b3cbdd4

balancerPoolId, qualifyingBalancerShares, err := k.prepareBalancerPoolAsFullRange(ctx, poolId, uptimeAccums)

osmosis/x/concentrated-liquidity/incentives.go

Line 441 in b3cbdd4

_, err := k.claimAndResetFullRangeBalancerPool(ctx, poolId, balancerPoolId, uptimeAccums)

We of course should make an effort to remove dead code and keep tests clean, so the following changes should also be made:

• Delete prepareBalancerPoolAsFullRange
• Delete claimAndResetFullRangeBalancerPool
• Remove all uses of reward splitting from tests
• If necessary/applicable, remove BalancerSharesRewardDiscount as a param as well as the corresponding default value (DefaultBalancerSharesDiscount) from types/constants.go

Acceptance Criteria

• All CL tests pass without issue
• All uses of "reward splitting" are removed from CL