break: Handle codes not in the dictionary

[robert3005]

In order to avoid performance penalty of checked accesses we always pad the symbols table to the max length. This will require a change in consumers to avoid serialising padded values

fixes #218

[codspeed-hq]

Merging this PR will not alter performance

✅ 30 untouched benchmarks

---

_{Comparing rk/handleoutofboundscodes (302e7cb) with develop (57969d6)}

[AdamGS]

This seems like a reasonable fix, if there's a way to make this faster we can do it as a followup.

[a10y]

If I try and decode with a symbol table and I encounter a code not in that table I should panic shouldn't I? But instead this will just proceed?

[robert3005]

not sure we should panic, we could offer a checked decompress that errors for symbols out of bounds?