drop the quickPlot dependency

[achubaty]

Closes #54.

quickPlot is being archived on CRAN and is no longer maintained. It was in Imports, so its archival would have taken NetLogoR with it. This removes it from DESCRIPTION entirely — nothing in R/, tests/, vignettes/ or inst/ references it any more.

What was actually coupled

NetLogoR only ever borrowed four generics and exported its own methods for them:

Generic	Was
coordinates	setGeneric("coordinates", quickPlot::coordinates)
extent	setGeneric("extent", quickPlot::extent)
layerNames	setGeneric("layerNames", quickPlot::layerNames)
numLayers	S3 methods registered on quickPlot's generic

These are now defined in NetLogoR with the same signatures, so calling code is unaffected. bbox() already worked this way, so there was a template in-tree.

The only piece that genuinely needed quickPlot was .identifyGrobToPlot(), which dispatches on quickPlot's internal .quickPlotGrob class so Plot() could draw a worldArray.

Why Plot() support could not be kept

quickPlot::Plot() dispatches on quickPlot's generics. Once the generics move into NetLogoR, keeping Plot() working means registering the methods against both sets — which puts two S4 method tables in the namespace and emits, at R CMD INSTALL and on every library(NetLogoR) for anyone with quickPlot installed:

Warning: multiple methods tables found for 'layerNames'
Warning: multiple methods tables found for 'extent'
Warning: multiple methods tables found for 'coordinates'

Verified causally by removing the dual registrations one at a time: 3 warnings with them, 0 without. Without the dual registration Plot() fails for all three classes (worldMatrix, worldArray, agentMatrix) at the extent dispatch, so the hook was unreachable dead code.

Plot() on NetLogoR objects therefore no longer works. plot() and points(), which draw through terra, are the replacement. Note this means no deprecation warning ships — there is no code path left to warn from, so quickPlot::Plot(w1) now surfaces quickPlot's own "unable to find an inherited method" error. The NEWS.md entry carries the explanation.

Also in here

• R/quickPlot.R → R/layers.R, since what remains is just the layer accessors.
• world2spatRast() passed nrow=/ncol= to terra::rast(), which partial-matched and warned on every plot of a world. Corrected to nrows=/ncols= — this clears 7 warnings that were present in every test run.
• test-plotting.R rewritten against plot()/points(); vignette prose updated in both .Rmds; all four shipped examples updated, including the SpaDES module's reqdPkgs.

Verification

• R CMD check --as-cran: 0 errors, 0 warnings, 0 notes
• devtools::test(): 0 failures (and down from 8 warnings to 1, itself pre-existing and unrelated)
• _R_CHECK_DEPENDS_ONLY_=true check: passes
• Smoke-tested against a real install with quickPlot not loaded: plot(), points(), numLayers(), layerNames(), extent() and coordinates() all behave, with zero "multiple methods tables" warnings

CI

All three workflows stop requesting PredictiveEcology/quickPlot@development in extra-packages, since nothing in the package references it any more. It can still arrive transitively as a dependency of the SpaDES packages; it is simply no longer something NetLogoR asks for.

🤖 Generated with Claude Code