Fix BLAS / LAPACK calls on x86

[willtebbutt]

Addresses #289 . Sadly, I think I need to add an 32 bit runner to ensure that this works correctly.

edit: a single x86 run has been added. Once CI has passed, I'll check that CI on the x86 runner does actually fail without the change made in this PR, before re-running CI and merging.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Files with missing lines	Coverage Δ
src/rrules/blas.jl	98.26% <100.00%> (+<0.01%)	⬆️

[github-actions]

Performance Ratio:
Ratio of time to compute gradient and time to compute function.
Warning: results are very approximate! See here for more context.

┌────────────────────────────┬──────────┬─────────┬─────────────┬─────────┐
│                      Label │ Mooncake │  Zygote │ ReverseDiff │  Enzyme │
│                     String │   String │  String │      String │  String │
├────────────────────────────┼──────────┼─────────┼─────────────┼─────────┤
│                   sum_1000 │     70.5 │     1.0 │         5.5 │ missing │
│                  _sum_1000 │     6.63 │  1200.0 │        34.3 │ missing │
│               sum_sin_1000 │     2.29 │    1.66 │        10.8 │ missing │
│              _sum_sin_1000 │     2.78 │   301.0 │        14.3 │ missing │
│                   kron_sum │     76.7 │    4.78 │       266.0 │ missing │
│              kron_view_sum │     66.5 │    10.4 │       265.0 │ missing │
│      naive_map_sin_cos_exp │     2.52 │ missing │        7.11 │ missing │
│            map_sin_cos_exp │     2.78 │    1.43 │         6.2 │ missing │
│      broadcast_sin_cos_exp │     2.68 │    2.28 │        1.46 │ missing │
│                 simple_mlp │     5.89 │    2.58 │        9.41 │ missing │
│                     gp_lml │     11.3 │    4.45 │     missing │ missing │
│ turing_broadcast_benchmark │     3.76 │ missing │        27.1 │ missing │
│         large_single_block │     4.13 │  4250.0 │        31.8 │ missing │
└────────────────────────────┴──────────┴─────────┴─────────────┴─────────┘