Add HitcountsIterableMapObserver, rename AsMutIter to AsIterMut

[domenukk]

This brings back iterators to HitcountsMapObserver.
I would be surprised if it were a lot slower, @tokatoka pleas check :)

[tokatoka]

I am against moving to iterators tbh...
it makes hitcounts slower
I ran forkserver_simple with introspection feature:
./forkserver_simple ./target/release/program corpus/

Current main build:

[PerfMonitor #0] run time: 0h-0m-15s, clients: 1, corpus: 3, objectives: 1, executions: 109072, exec/sec: 7268
Client 000:
  0.0001: Scheduler
  0.0000: Manager
  Stage 0:
    0.0001: GetInputFromCorpus
    0.0059: Mutate
    0.0000: MutatePostExec
    0.7372: TargetExecution
    0.0045: PreExecObservers
    0.0893: PostExecObservers
  Feedbacks:
    0.1598: mapfeedback_metadata_shared_mem
    0.0002: time
    0.0001: CrashFeedback
  0.0027: Not Measured

[PerfMonitor #0] run time: 0h-0m-30s, clients: 1, corpus: 3, objectives: 1, executions: 218192, exec/sec: 7267
Client 000:
  0.0001: Scheduler
  0.0000: Manager
  Stage 0:
    0.0001: GetInputFromCorpus
    0.0059: Mutate
    0.0000: MutatePostExec
    0.7371: TargetExecution
    0.0045: PreExecObservers
    0.0895: PostExecObservers
  Feedbacks:
    0.1597: mapfeedback_metadata_shared_mem
    0.0002: time
    0.0001: CrashFeedback
  0.0027: Not Measured

^C

iter branch build:

[PerfMonitor #0] run time: 0h-0m-15s, clients: 1, corpus: 3, objectives: 1, executions: 93914, exec/sec: 6255
Client 000:
  0.0003: Scheduler
  0.0000: Manager
  Stage 0:
    0.0002: GetInputFromCorpus
    0.0059: Mutate
    0.0000: MutatePostExec
    0.6957: TargetExecution
    0.0040: PreExecObservers
    0.1477: PostExecObservers
  Feedbacks:
    0.0001: time
    0.0001: CrashFeedback
    0.1437: mapfeedback_metadata_shared_mem
  0.0022: Not Measured

[PerfMonitor #0] run time: 0h-0m-30s, clients: 1, corpus: 3, objectives: 1, executions: 188402, exec/sec: 6273
Client 000:
  0.0003: Scheduler
  0.0000: Manager
  Stage 0:
    0.0002: GetInputFromCorpus
    0.0059: Mutate
    0.0000: MutatePostExec
    0.6954: TargetExecution
    0.0040: PreExecObservers
    0.1479: PostExecObservers
  Feedbacks:
    0.0001: time
    0.0001: CrashFeedback
    0.1438: mapfeedback_metadata_shared_mem
  0.0022: Not Measured

it has nearly 1.6x the time spent for the postexec i.e. hitcountsobserver

[domenukk]

If the MultiMap has odd-sized maps, this will break/be unsafe

[codecov-commenter]

Codecov Report

Merging #713 (7da0dbc) into main (a5248d0) will decrease coverage by 0.03%.
The diff coverage is 10.29%.

@@            Coverage Diff             @@
##             main     #713      +/-   ##
==========================================
- Coverage   15.02%   14.99%   -0.04%     
==========================================
  Files         146      146              
  Lines       17336    17377      +41     
==========================================
  Hits         2605     2605              
- Misses      14731    14772      +41     

Impacted Files	Coverage Δ
libafl/src/bolts/mod.rs	16.66% <ø> (ø)
libafl/src/feedbacks/map.rs	13.12% <0.00%> (+0.17%)	⬆️
libafl/src/stages/calibrate.rs	0.00% <ø> (ø)
libafl/src/observers/map.rs	4.04% <10.60%> (-0.40%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a5248d0...7da0dbc. Read the comment docs.

[tokatoka]

How about making another MultiMapHitcountsObserver that can handle multiple maps

[domenukk]

Ooh, or we could have one hitcount observer per sub-map