test: balance sanitization

[sebastiantf]

Description

• Part of exploring @flyingnobita's lead on balance sanitization
• Added a few tests that tries providing negative values to csv_parser, mst, range_check, big_uint_to_fp conversions

Observations

1. Cannot convert negative number into BigUint: test_negative_big_uint_conversion

• BigUint conversion using to_biguint returns None. So unwrap() panics
• Hence cannot convert negative number to big uint
• BigUint is being converted to field element using merkle_sum_tree::utils::big_uint_to_fp

2. Field elements are non-negative values. So it doesn't seem possible to get a negative value into mst or range_check: test

• halo2_proofs::halo2curves::bn256::Fr does not implement From trait for any signed integer types
• Sidenote: Summa seems to use their own fork of halo2 which seems to use an old version of halo2curves v0.1.0 whereas PSE's halo2curves is at v0.6.1

3. csv_parser and merkle_sum_tree errors when using negative values: test_parse_csv_to_entries, test_mst_negative_value_in_csv

• This is a direct effect of Observation 1 from the line

[0xnirlin]

Regarding this, one thing I was thinking, about is it even possible for an individual user of an exchange to have a negative value. I don't think it is possible.

[flyingnobita]

Yeah it "shouldn't" be possible. I believe that's why we need to put these constraints in the code to prevent a malicious custodian from undermining the system in anyway possible.

[igorline]

Same, haven't found yet any way to submit invalid value into the circuit as BigUint used won't allow this