Feature/summary

[rogerssam]

Add summary() for design objects (and a leaner print())

What

Adds a summary.design() method (with a print.summary.design()) that turns a design object into a statistically meaningful evaluation. print() gives a snapshot of main optimisation results, and summary() lets you evaluate and defend it. Targets feature/summary; bumps to 0.0.9.

Why

The single combined score that print() showed isn't enough to judge a design. Biometricians (and workshop audiences) need to see replication, how the score breaks down, and design-quality diagnostics like connectedness and concurrence without dropping into the object internals by hand.

Changes

summary.design — returns a queryable "summary.design" list (printing is formatting-only) in three sections:

• Structure — layout, per-level treatments/replication, spatial factors and their levels.
• Optimisation — objective used, the score decomposed into its additive components (adjacency/balance for the default objective; neighbour-balance/even-distribution/balance/adjacency for Piepho; main/interaction for factorial), initial → final, iterations and temperature schedule.
• Evaluation — connectedness, within-block concurrence, replicate spatial span, opt-in A-efficiency, and neighbour balance. Plus prominent Flags (didn't converge, disconnected, unequal replication) when relevant.

Self-describing design objects — speed() results now carry a metadata field recording each level's swap variable, spatial factors, weights, optimisation settings, achieved score and score components. This is the prerequisite that lets summary() report faithfully rather than re-derive.

Key design decisions

• Score components come from the optimisation run, not a recompute. Each built-in objective now returns a components vector (the additive pieces that sum to its score), captured into metadata with the exact arguments used. So the decomposition is always faithful — including non-additive objectives (Piepho/factorial) and ring-weighted adjacency — and custom objectives that don't expose components simply show the score with no decomposition.
• Connectedness is a model-based estimability check (lm(~ spatial factors + block + treatment), counting aliased treatment contrasts), base R, no lme4. It adjusts for whatever the design is stratified by, so incomplete-block designs are assessed the way they're actually analysed. It is size-guarded: skipped for very large designs (where the dense fit is
  expensive) unless forced.
• Concurrence is reported only for incomplete blocks. For complete designs (RCBD, split-plot) every concurrence equals the replication and carries no information, so it is auto-skipped (overridable).
• A-efficiency is opt-in (efficiency = TRUE) — it is the heaviest metric and assumes a row–column model.
  • This needs to be revisited after the A-optimality objective is merged in.
• Neighbour balance auto-enables when the design was optimised with the Piepho objective.
• Buffers are excluded. Plots added by add_buffers() are a field-layout convenience, not part of the statistical design, so print() and summary() silently ignore them.
• No new dependencies — everything is base R.

Tests / docs

New test-summary.R plus additions across the objective/efficiency tests; roxygen and runnable examples for the new methods; a new "Evaluating a Design" vignette section; a NEWS.md entry. Full test suite green.