Add interface for DBSCAN

[ablaom]

Replaces #11. Builds on JuliaAI/MLJBase.jl#806, which was not available at that time.

Adds an interface for the DBSCAB model, implementing it as a Static model (no fit), as the model does not generalize to new data. Assignments are returned by predict; DBSCAN point types (core/boundary/noise) are accessed from the post-predict report. Detailed MLJ-compliant doc-string included.

using MLJBase
using MLJClusteringInterface

X, y  = make_moons(400, noise=0.09, rng=1)

model = DBSCAN(radius=0.13, min_cluster_size=5)
mach = machine(model)

# compute and output cluster assignments for observations in `X`:
yhat = predict(mach, X);

# compare with true labels:
compare = zip(yhat, y) |> collect;
julia> compare[1:10] # clusters align with classes
10-element Vector{Tuple{CategoricalArrays.CategoricalValue{Int64, UInt32}, CategoricalArrays.CategoricalValue{Int64, UInt32}}}:
 (1, 1)
 (2, 0)
 (2, 0)
 (1, 1)
 (2, 0)
 (2, 0)
 (2, 0)
 (2, 0)
 (2, 0)
 (2, 0)

# access post-predict report (`C` is core point, `N` is noise):
julia> unique(report(mach).point_types)
2-element Vector{Char}:
 'C': ASCII/Unicode U+0043 (category Lu: Letter, uppercase)
 'N': ASCII/Unicode U+004E (category Lu: Letter, uppercase)

# other report items:
julia> report(mach) |> keys
(:point_types, :nclusters, :cluster_labels, :clusters)

[ablaom]

@juliohm

[codecov-commenter]

Codecov Report

Merging #17 (85cfa5c) into master (867f7f0) will increase coverage by 3.67%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #17      +/-   ##
==========================================
+ Coverage   94.73%   98.41%   +3.67%     
==========================================
  Files           1        1              
  Lines          38       63      +25     
==========================================
+ Hits           36       62      +26     
+ Misses          2        1       -1     

Impacted Files	Coverage Δ
src/MLJClusteringInterface.jl	98.41% <100.00%> (+3.67%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[OkonSamuel]

LGTM!!!.

[ablaom]

Thanks @OkonSamuel for your review.