feat: add special asset pairs

[toteki]

Contains logic for creating and storing these pairs, but not their actual effects on borrow limit.

These can be released early to accelerate their proposal timeline.

[codecov]

Codecov Report

Merging #2150 (5e2e948) into main (7f05ad4) will decrease coverage by 6.52%.
Report is 160 commits behind head on main.
The diff coverage is 82.35%.

@@            Coverage Diff             @@
##             main    #2150      +/-   ##
==========================================
- Coverage   75.38%   68.86%   -6.52%     
==========================================
  Files         100      151      +51     
  Lines        8025    11680    +3655     
==========================================
+ Hits         6050     8044    +1994     
- Misses       1589     3095    +1506     
- Partials      386      541     +155     

Files Changed	Coverage Δ
ante/spam_prevention.go	75.92% <ø> (ø)
x/incentive/codec.go	47.82% <ø> (+9.89%)	⬆️
x/incentive/keeper/invariants.go	0.00% <0.00%> (ø)
x/incentive/keeper/store.go	70.67% <ø> (ø)
x/incentive/keeper/unbonding.go	80.45% <ø> (ø)
x/incentive/keeper/update.go	52.11% <ø> (ø)
x/incentive/keys.go	100.00% <ø> (ø)
x/incentive/msgs.go	77.41% <ø> (-0.81%)	⬇️
x/incentive/params.go	89.28% <ø> (-10.72%)	⬇️
x/leverage/client/tests/suite.go	100.00% <ø> (ø)
... and 55 more

... and 56 files with indirect coverage changes

[robert-zaremba]

• need to remove title and description from gov messages
• let's talk about using list of assets rather than a pair.
• use store helpers

[kosegor]

Overall looks good, adding some minor comments.

[robert-zaremba]

• Instead of simplifying user interface by using sets, this iteration added additional complexity to add both pairs and sets. Moreover the the pars are stored in each direction. So for a set (A,B,C) we store 6 pairs (and return 6 pairs when querying it, instead of 1): (A,B), (A,C), (B,C), (B,A), (B,C), (C,B)` -- that is against what we wanted based on the discussion in Slack.
• if under the hood we save sets as pairs, then we need to rethink the key structure and assure we don't have duplicates.
• let's don't create new error codes, which are already deprecated.
• prefer to not use yaml
• let's make sure we have unit-tests task in the Linear.

merging this as discussed