[EPIC] shrex stability + performance

[walldiss]

Issues

### Shrex-nd:
- [x] https://github.com/celestiaorg/celestia-node/issues/2035
- [ ] https://github.com/celestiaorg/celestia-node/issues/2044
- [ ] https://github.com/celestiaorg/celestia-node/issues/2051
- [ ] https://github.com/celestiaorg/celestia-node/issues/2066
- [ ] https://github.com/celestiaorg/celestia-node/issues/2089
- [ ] https://github.com/celestiaorg/celestia-node/issues/2129
- [ ] https://github.com/celestiaorg/celestia-node/issues/2145
- [ ] https://github.com/celestiaorg/celestia-node/issues/2174
- [ ] https://github.com/celestiaorg/celestia-node/issues/2183
- [ ] https://github.com/celestiaorg/celestia-node/issues/2183

### Shrex-eds:
- [ ] https://github.com/celestiaorg/celestia-node/issues/2077
- [ ] https://github.com/celestiaorg/celestia-node/issues/2109
- [ ] https://github.com/celestiaorg/celestia-node/issues/2110
- [ ] https://github.com/celestiaorg/celestia-node/issues/2176
- [ ] https://github.com/celestiaorg/celestia-node/issues/2146

### Peer Manager + Discovery
- [ ] https://github.com/celestiaorg/celestia-node/issues/2107
- [ ] https://github.com/celestiaorg/celestia-node/pull/1924
- [ ] implement client side per peer rate-limiting in peer-manager
- [ ] implement jitter to peer cooldown in peer-manager
- [ ] https://github.com/celestiaorg/celestia-node/pull/1924
- [ ] bug(shrex/peer-manager): full nodes are given their own pool after `ResultCooldownPeer`
+ [ ] https://github.com/celestiaorg/celestia-node/issues/2152

### Non-critical:
 - [ ] https://github.com/celestiaorg/celestia-node/issues/1997
 - [ ] https://github.com/celestiaorg/celestia-node/issues/2016

### Action plan

- [ ] #2143 
- [ ] https://github.com/celestiaorg/celestia-node/pull/2101

  ## Performance
  -  [ ] https://github.com/celestiaorg/celestia-node/pull/2101
  - [ ] duplicate messages from shrex.Sub
  - [ ] replace floodSub in shrex.Sub by celestia.Messanger