engine: a column's \end{minipage} arrives through \column and \end{columns}

[tannevaled]

beamer opens a column's minipage and defers its \end into \beamer@colclose:

\newcommand<>\beamer@columncom[2][\beamer@colmode]{%
  \beamer@colclose
  \def\beamer@colclose{\end{minipage}\hfill\end{actionenv}\ignorespaces}%
  \begin{actionenv}#3…\begin{minipage}…}
                          % beamerbaseframecomponents.sty:281-283

so the \end this scan hunts arrives through the next \column, or through \end{columns} (:237 — \endcolumns begins with \beamer@colclose). A raw scan sees neither: \column takes arguments where the narrow expandsToEnd rule only reaches parameterless macros, and \end{columns} is stored as somebody else's \end.

Read raw, the first column swallowed its sibling and everything after the frame. A talk with two columns in a [fragile] frame rendered one page of its seven. Issue #127 found this; #129 caught only the case where the frame's own file ends there, which is why it came back.

Both are now read as what they are:

• at \column, the body ends and \beamer@colclose is emptied — its \end{minipage} has just been honoured, which is what beamer itself does after running it (:269, \beamer@colclose\def\beamer@colclose{});
• at \end{columns}, \end<columns>'s body is pushed so \beamer@colclose, and then its \end{minipage}, surface for the depth bookkeeping.

Measured — 200 beamer talks, 200 arXiv papers

5 talks change, two closer to tectonic and none further:

	avant	après	tectonic
	1	5	7
	2	4	4

beamer gains +2885 glyphs and its page error against the 80 tectonic renders falls 48 → 42, with one more exact match (62/80). arXiv does not move by one glyph.

Test

columnclosesminipage_test.go: two columns in a [fragile] frame, with a frame on either side — three pages, and the last frame's text must be on the page. On main it is one page carrying only the first frame.