Compress the serialized PCD collection during sync if above a threshold size

[robknight]

This PR amends the serializeStorage and deserializeStorage to apply gzip compression to the serialized PCD string. By pre-compressing the string, we reduce the size of the plaintext sync blob. This is helpful as compression cannot be applied after the blob has been encrypted.

Testing with pseudo-random data indicates a reduction of 50% in the size of the PCD collection string. We might hope to see better real-world performance as real data will have more repetition (of names, email addresses, public keys, and so on).

[rrrliu]

this looks like a pretty big package for our bundle (>1MB) -- any way we could dynamically load via bundle splitting or perhaps use a smaller package?

https://npm-compare.com/fflate,pako

https://packagephobia.com/result?p=pako%402.1.0

[robknight]

It's already a dependency for passport-ui, where it's used to compress the contents of QR codes (at least in cases where those QR codes contain large data, e.g. proofs). It takes up 46kb of space in the bundle:

Since it's already included by passport-ui, the additional cost of using it here is zero.

I think packagephobia is just counting the total size of the package's dist directory, but only one file from that directory is actually used - probably pako.min.js or pako.es5.min.js, both of which are around 46kb in size.