noir-lang/big_curve test test_secp256r1_offset_generators fails after #6685

[TomAFrench]

While fixing the dead canary @kashbrti found in noir-lang/noir_bigcurve#22 that the test_secp256r1_offset_generators would fail with the error

error: Failed to solve program: 'Failed to solve brillig function'
   ┌─ /mnt/user-data/tom/nargo/github.com/noir-lang/noir-bignumv0.4.2/src/fns/expressions.nr:74:12
   │
74 │     assert(remainder == [0; N]);
   │            -------------------
   │
   = Call stack:
     1. src/bigcurve_test.nr:534:17
     2. src/lib.nr:331:9
     3. src/lib.nr:772:13
     4. src/lib.nr:380:9
     5. /mnt/user-data/tom/nargo/github.com/noir-lang/noir-bignumv0.4.2/src/bignum.nr:264:9
     6. /mnt/user-data/tom/nargo/github.com/noir-lang/noir-bignumv0.4.2/src/fns/expressions.nr:279:9
     7. /mnt/user-data/tom/nargo/github.com/noir-lang/noir-bignumv0.4.2/src/fns/expressions.nr:74:12

I've bisected this down to passing on fc11b63 and failing on 24cc19e so it seems like the root cause is due to us not copying arrays when we should due to the changes made in #6685.

We didn't catch this at the time as the tests in noir_bignum were already failing in CI due to being too heavy for the github runner to run in parallel.

[TomAFrench]

@jfecher Can you take a look at whether we can salvage #6685 or if it needs to be reverted?

[jfecher]

Looking into this now