Fix single-sample crash in plotSubclonePie (Columns 2 and 3 must be named) (#53)

[alexanderchang1]

Closes #53.

Problem

A single-sample (S=1) runPictograph() completes JAGS sampling and writes all core outputs (tree.csv, mcf.csv, mutationClusterAssign.csv, subclone_proportion.csv), then aborts drawing the pie chart:

Error in `as_tibble()`: ! Columns 2 and 3 must be named.
Calls: runPictograph -> print -> plotSubclonePie -> ... -> as_tibble

Because the error propagates out of runPictograph(), inference has already succeeded but downstream steps (e.g. runDeconvolution) are skipped. Multi-sample runs are unaffected.

Root cause

In runPictograph, subclone_props is reordered with matrix subsetting:

subclone_props <- subclone_props[order(as.numeric(rownames(subclone_props))), ]

For a K×1 matrix this drops the dimension (default drop=TRUE) and collapses it to a length-K vector. plotSubclonePie() then takes the is.null(dim(...)) branch and as_tibble_row() produces a 1-row K-column tibble, but sample_names has length 1, so the columns come out unnamed and as_tibble() errors.

Fix

• Preserve the matrix dimension at the source with drop = FALSE at both reorder sites in runPictograph (the single-tree and per-tree paths).
• Add a defensive guard at the top of plotSubclonePie() that restores a dropped dimension, so the function is robust to a vector argument from any caller.

Validated on single-sample PDO runs that previously aborted at the pie chart.