Fix rule extraction to handle constant factors

[Zapiano]

Note: This branch is based on #863 's branch. Closes #672 .

---

Updates ADRIA.analysis.cluster_rules so we can check for constant factors (and ignore them) before trying to extract rules. The parameters you need to pass to cluster_rules have changed - with that is one line less of code now.

The documentation still needs to be updated - once this is done, I'll mark this as ready to review.

Example

As an example, let's fix all SeedCriteriaWeights and only a few of the FogCriteriaWeights factors, run the model, cluster and target some cluster:

dom = ADRIA.load_domain("...")

ADRIA.fix_factor!(
    dom;
    # Seed criteria
    seed_out_connectivity=0.9,
    seed_heat_stress=0.8,
    seed_depth=0.7,
    seed_coral_cover=0.5,
    seed_in_connectivity=0.4,
    seed_wave_stress=0.0,
    # Fog criteria
    fog_coral_cover=0.5,
    fog_depth=0.7
)

# Create scenarios
num_scens = 2^8
scens = ADRIA.sample(dom, num_scens)

# Run model / Load ResultSet
rs = ADRIA.run_scenarios(dom, scens, "45")

# Find Time Series Clusters
s_tac = ADRIA.metrics.scenario_total_cover(rs)
num_clusters = 6
clusters = ADRIA.analysis.cluster_scenarios(s_tac, num_clusters)

# Target scenarios
target_clusters = ADRIA.analysis.target_clusters(clusters, s_tac)

Now if we select Intervention factors, everything works (as it would previously, nothing's c hanged here):

max_rules = 10
fields_iv = ADRIA.component_params(rs, [Intervention]).fieldname
rules_iv = ADRIA.analysis.cluster_rules(dom, target_clusters, scens, fields_iv, max_rules)

If we select FogCriteriaWeights, some are constant and some are not. Previously this would lead to an error but now it just filters off all constant factors and everything works fine:

max_rules = 10
fields_fog = ADRIA.component_params(rs, [FogCriteriaWeights]).fieldname
rules_fog = ADRIA.analysis.cluster_rules(
    dom, target_clusters, scens, fields_seeds, max_rules
)

And, finally, if we select SeedCriteriaWeights, because all of them are constant, this will raise an error:

max_rules = 10
fields_seeds = ADRIA.component_params(rs, [SeedCriteriaWeights]).fieldname
rules_seeds = ADRIA.analysis.cluster_rules(
    dom, target_clusters, scens, fields_seeds, max_rules
)

The error:

ERROR: ArgumentError: Factors of interest have to be non constant.
Stacktrace:
 [1] cluster_rules(domain::ADRIADomain, clusters::Vector{…}, scenarios::DataFrames.DataFrame, factors::Vector{…}, max_rules::Int64; seed::Int64, kwargs::@Kwargs{})
   @ ADRIA.analysis c:\Users\pribeiro\AIMS\Code\ADRIA.jl\src\analysis\rule_extraction.jl:148
 [2] cluster_rules(domain::ADRIADomain, clusters::Vector{…}, scenarios::DataFrames.DataFrame, factors::Vector{…}, max_rules::Int64)
   @ ADRIA.analysis c:\Users\pribeiro\AIMS\Code\ADRIA.jl\src\analysis\rule_extraction.jl:134
 [3] top-level scope
   @ c:\Users\pribeiro\AIMS\Code\ADRIA.jl\sandbox\rule_extraction\fixed_factors.jl:77
Some type information was truncated. Use `show(err)` to see complete types.

Below are the scatter plots for Interventions and FogCriteriaWeights rules (the plot functions haven't changed):

Intervention	FogCriteriaWeights

[ConnectedSystems]

Haven't looked through the changes yet (will wait for the dependent PR to be merged first), but one minor request:

Change the error message to "Factors of interest cannot be constant".

[Zapiano]

@ConnectedSystems this one is ready for review.

[Zapiano]

@ConnectedSystems friendly reminder that this is ready for review

[ConnectedSystems]

Minor comment - could this be simplified to ResultSet? If not, maybe we should consider exporting the ResultSet type...

[Zapiano]

Yep, that works.

[ConnectedSystems]

Just one minor question.

Otherwise this is ready to be merged.

[Zapiano]

I still need to rebase this.

[Zapiano]

@ConnectedSystems this one is rebased and ready for review.