Random Padding in P2P Messages

## Overview

From https://github.com/handshake-org/hsd/pull/287

```
* A future possibility may be to randomly pad out messages
* to avoid suspicion from a tyrannical government
* (todo: research/discuss this more).
```

Opening this issue to begin discussion.

Randomly padding the p2p messages is a bigger rabbit hole than I expected. Wikipedia has some good primers on the topic:

- https://en.wikipedia.org/wiki/Padding_(cryptography)#Traffic_analysis_and_protection_via_padding
- https://en.wikipedia.org/wiki/Traffic_analysis
- https://en.wikipedia.org/wiki/PURB_(cryptography)

There are two main types of padding schemes:
- Deterministic
- Places packets into buckets, padding each packet in a bucket to the same size
- Must choose bucket cutoffs wisely, many messages around the cutoff can leak information
- Can pad to a common size or to the next power of 2
- [Padme, Section 4](https://bford.info/pub/sec/purb.pdf)/[PURBS](https://en.wikipedia.org/wiki/PURB_(cryptography)) seem promising
- Random
- Statistical techniques can be used to deanonymize packets over time
- If random pad size is smaller than the packet, it doesn't give strong anonymity benefits over time
- If random pad size is larger than the packet, there is a large overhead

To account for the overhead that comes with padding all p2p messages with random data, an adapted version of Erlay could be used:
- https://github.com/naumenkogs/bips/blob/bip-reconcil/bip-reconcil.mediawiki
- https://arxiv.org/abs/1905.10518

## Implementation

- Could be added to `net` package or `brontide` package
- I think its a better idea to add it to the `brontide` package since that layer is responsible for encryption
- Does this block mainnet, or could it be soft forked in?