feat: support loading stable memory from a file

[ielashi]

Problem

Running benchmarks that require a particular stable memory is currently not possible with canbench. I ran into this problem quite often when wanting to benchmark AI models, and I'm sure there are other use-cases for it as well.

Solution

This PR introduces a way to specify a binary file in canbench.yml that will be loaded as the canister's stable memory after initialization.

Example:

stable_memory:
  file:
    stable_memory.bin

[dsarlis]

I think this also suffers from the same issue as #65. Namely that some benchmark is giving different results on different platforms.

[ielashi]

I think this also suffers from the same issue as #65. Namely that some benchmark is giving different results on different platforms.

I addressed by making the test only ensure success. We don't actually need to assert on the exact instruction count, as that is quite brittle.

[github-actions]

canbench 🏋 (dir: examples/fibonacci)

No significant performance changes detected ✅

examples/fibonacci/canbench_results.yml is up to date ✅

---------------------------------------------------

Benchmark: fibonacci_20
  total:
    instructions: 2224 (no change)
    heap_increase: 0 pages (no change)
    stable_memory_increase: 0 pages (no change)

---------------------------------------------------

Benchmark: fibonacci_45
  total:
    instructions: 3011 (no change)
    heap_increase: 0 pages (no change)
    stable_memory_increase: 0 pages (no change)

---------------------------------------------------

[github-actions]

canbench 🏋 (dir: examples/btreemap_vs_hashmap)

No significant performance changes detected ✅

examples/btreemap_vs_hashmap/canbench_results.yml is up to date ✅

---------------------------------------------------

Benchmark: insert_users
  total:
    instructions: 2.57 B (no change)
    heap_increase: 871 pages (no change)
    stable_memory_increase: 0 pages (no change)

---------------------------------------------------

Benchmark: remove_users
  total:
    instructions: 2.07 B (no change)
    heap_increase: 0 pages (no change)
    stable_memory_increase: 0 pages (no change)

---------------------------------------------------

Benchmark: pre_upgrade_bench
  total:
    instructions: 743.98 M (0.00%) (change within noise threshold)
    heap_increase: 519 pages (no change)
    stable_memory_increase: 184 pages (no change)

  serialize_state (scope):
    instructions: 729.05 M (0.00%) (change within noise threshold)
    heap_increase: 519 pages (no change)
    stable_memory_increase: 0 pages (no change)

  writing_to_stable_memory (scope):
    instructions: 14.93 M (no change)
    heap_increase: 0 pages (no change)
    stable_memory_increase: 184 pages (no change)

---------------------------------------------------