v0.6.2

New examples

• Multiple-pass estimation (gsjs/example_gsjs_multipass) — the Harding & Deutsch
  variance-calibrated MPE workflow (geostatisticslessons.com/lessons/multipass), end to end:
  ground truth is a real unconditional SGSIM realization (@atra/gslib, Wasm) from a known
  nested variogram, drilled at three spacings; each kriging pass is calibrated into the
  change-of-support f window against a nearest-neighbour reference; SPLS and NN straw men
  bracket the calibrated model in a grade–tonnage verdict judged against the simulated truth.
  All parameters are live sliders.
• Multiple-pass estimation, Python edition (gsjs/example_gsjs_multipass_py) — the same
  document with every cell in adder (Python). The RNG is ported bit-for-bit, so both editions
  produce identical numbers — two languages, one verifiable result.

@gcu/plot

• Perceptual colormaps (viridis, coolwarm, turbo, plasma, inferno, magma, cividis) now
  interpolate real anchor stops instead of polynomial fits that went badly wrong at the ends
  (viridis topped out green instead of yellow; coolwarm ended white instead of dark red).
• cmap accepts a list of colors as a categorical map (matplotlib's ListedColormap idiom).

micro

• Analysis-window footers: opaque --panel background, docked to the window bottom.