pointerset + friends with pointers-to-pointers

[willtebbutt]

Will address #470 .
Todo:

• add failing primitive tests for pointerset and atomic_pointerset
• add failing derived tests for pointerset and atomic_pointerset (to enable testing of numerical correctness)
• fix problem
• bump patch version

edit: it looks like pointerref doesn't suffer from this problem, which is a relief.

[codecov]

Codecov Report

Attention: Patch coverage is 4.34783% with 22 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/rrules/builtins.jl	4.34%	22 Missing ⚠️

Files with missing lines	Coverage Δ
src/rrules/builtins.jl	48.70% <4.34%> (-50.27%)	⬇️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[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 │   Primal │ Mooncake │  Zygote │ ReverseDiff │  Enzyme │
│                     String │   String │   String │  String │      String │  String │
├────────────────────────────┼──────────┼──────────┼─────────┼─────────────┼─────────┤
│                   sum_1000 │ 100.0 ns │      1.8 │     1.1 │         5.5 │    8.11 │
│                  _sum_1000 │ 941.0 ns │     6.99 │  1700.0 │        33.5 │    1.07 │
│               sum_sin_1000 │  6.56 μs │     2.21 │    1.68 │        10.5 │    2.21 │
│              _sum_sin_1000 │  5.29 μs │      2.6 │   314.0 │        13.0 │    2.46 │
│                   kron_sum │ 349.0 μs │     37.0 │    4.79 │       202.0 │    11.0 │
│              kron_view_sum │ 348.0 μs │     41.1 │    10.4 │       240.0 │    9.31 │
│      naive_map_sin_cos_exp │  2.15 μs │     2.18 │ missing │        7.17 │    2.33 │
│            map_sin_cos_exp │  2.11 μs │     2.44 │     1.5 │        6.13 │     2.9 │
│      broadcast_sin_cos_exp │  2.22 μs │      2.3 │    2.39 │        1.54 │    2.27 │
│                 simple_mlp │ 412.0 μs │      4.7 │    1.55 │        6.76 │    3.14 │
│                     gp_lml │ 537.0 μs │     4.59 │    2.41 │     missing │    3.13 │
│ turing_broadcast_benchmark │  1.98 ms │      3.5 │ missing │        25.1 │ missing │
│         large_single_block │ 390.0 ns │     4.39 │  4370.0 │        30.2 │    2.18 │
└────────────────────────────┴──────────┴──────────┴─────────┴─────────────┴─────────┘

[yebai]

@willtebbutt, can you address this issue to unblock DynamicPPL work?

[willtebbutt]

Ah, sorry, I forgot / didn't realise that this was blocking. The issues here are mainly to do with our testing infrastructure's handling of pointers being a little tricky to work with. I need to sort that out before this can be safely merged -- I think I've figured out what needs doing while working on the forwards-mode PR, so I'll take a look this afternoon.