fix: use protons and update project config

[achingbrain]

• Swaps protobufjs for protons so we can generate .ts from .proto that is browser friendly (e.g. uses bigints instead of long.js, etc) and isn't broken by default
• Updates project config to add autorelease in the unified CI model as requested in this comment - fix: access node process object via globalThis #136 (comment)

Before merging this:

1. Check npm config of https://www.npmjs.com/package/@chainsafe/libp2p-noise - "Publishing access" should be set to "Require two-factor authentication or automation tokens"
2. Create a repository secret here called NPM_TOKEN with a npm automation token that can publish to @chainsafe/libp2p-noise.
3. Add https://github.com/web3-bot as a collaborator to this repo to receive automated CI config updates

[achingbrain]

Yarn.

So, for some reason, yarn@1.22.18 consistently symlinks the wrong tsc binary in node_modules/.bin.

The dep tree is:

@chainsafe/libp2p-noise@6.0.1 /Users/alex/Documents/Workspaces/ChainSafe/js-libp2p-noise
└─┬ aegir@37.0.12
.. other deps
  ├─┬ dependency-check@5.0.0-4
  │ └─┬ precinct@8.3.1
  │   └─┬ detective-typescript@7.0.2
  │     └── typescript@3.9.10
.. other deps
 └── typescript@4.6.3

It finds typescript@3.9.10 as a dep of detective-typescript instead of typescript@4.6.3 as a dep of aegir.

This is a problem because typescript@3.9.10 doesn't support the right values for the lib option and compiling the typescript code fails.

One solution might be to add a redundant typescript dep to this module and hope that the dependency resolution algorithm would choose that one to symlink, though it's a bad solution because it's not used directly here, plus you'd then need to keep it in sync with the aegir version which is extra work no-one needs.

Unified CI uses npm and not yarn, so I've removed yarn.lock since it's redundant from a CI point of view but there's nothing to stop people using yarn if they want to, though at the moment it seems to have a bug in it.

[achingbrain]

I have split this into two PRs - #141 and #142