Optimize account token cache design

[aditya1702]

What

Optimizes Redis storage for the Account Token Cache to significantly reduce memory usage through multiple compression strategies.

1. PostgreSQL Asset ID Mapping - Store compact integer IDs in Redis instead of full CODE:ISSUER strings. Asset definitions are stored once in a new trustline_assets PostgreSQL table.
2. Bucket Sharding - Distribute accounts across 12,000 Redis hash buckets using FNV-1a hashing instead of one key per account. This reduces key overhead and improves Redis memory efficiency.
3. VARINT Binary Encoding - Encode asset IDs using variable-length integer format instead of ASCII
4. Frequency-Based ID Assignment - Sort assets by frequency during initial population so common assets (USDC, EURC) get lowest IDs (1, 2, 3...), which encode to fewer bytes in VARINT format.

With all these optimizations the memory usage went from 3 GB -> 387 MB.

Why

Optimizing memory usage for our token cache is important since the account keys will have an unbounded growth and this could lead to increased storage.

Known limitations

N/A

Issue that this PR addresses

Closes #422

[aristidesstaffieri]

maybe this should say a configurable size for Redis hash buckets or something to that effect?

[aristidesstaffieri]

would operators want to tune these as well, since they can tune the bucket size?

[aditya1702]

@aristidesstaffieri Yes I have added a section in README explaining how these can be tuned and affect the token cache

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	github.com/​dgraph-io/​ristretto/​v2@​v2.3.0

View full report