feat: Bye bye bcrypto

[acolytec3]

This is a redo of #197 and fixes #301.

Problem: discv5 uses bcrypto dependency for performant secp256k1 and sha256 functions. This library causes occasional CI failures in ultralight due to in explicable OOMs and other indecipherable C++ errors

Solution: Replace bcrypto usage with a discv5Crypto interface similar to the v4Crypto interface exposed by the enr package. As currently stands, this PR implements a defaultCrypto object using noble/hashes and noble/secp256k1 that are pure javascript implementation and are somewhat slower than the bcrypto equivalents but can easily be replaced with WASM compiled alternatives such as as-sha256 or polkadot/wasm-crypto.

Sidequest 1: Replace bcrypto AES cipher functionality with nodejs builtins that are overall 5-10% faster.
Sidequest 2: Replace bcrypto randomBytes function with noble equivalent

[acolytec3]

@wemeetagain Here's the initial draft. I haven't implemented the wasm-crypto alternative to noble yet (that is much more performant), but this basically what I'm thinking.

[acolytec3]

Note, the CI is failing because of a quirk in noble/hashes related to Node 18 and older versions. Would you be open to just bumping the nodejs version to 20 in CI to eliminate this issue? If not, I'll do the required shim but seems kind of silly given node 18 is EOL at this point.

[acolytec3]

Note, the CI is failing because of a quirk in noble/hashes related to Node 18 and older versions. Would you be open to just bumping the nodejs version to 20 in CI to eliminate this issue? If not, I'll do the required shim but seems kind of silly given node 18 is EOL at this point.

@wemeetagain would you be open to bumping CI to run on node 20 or do you want me to apply the shim that noble says is required to run on node 18? Note, ultralight is running on node 20 and above only so not an issue for us.

[acolytec3]

Other than the CI nodejs environment question, this should be ready for review.