CSV merge for text templates (one row per copy or per job) #528

Is your feature request related to a problem? Please describe.
Name tags, product labels and personalised gifts come with a list: a spreadsheet of names, article numbers or prices. Today every line has to be typed into a text box by hand, one job at a time. LightBurn calls this Variable Text "Merge/CSV".

Describe the solution you'd like
Building on the serial counter and variable offsets from #527:

• Link a CSV file to the project. The rows are stored in the project, so it stays self-contained; the file path is kept and can be reloaded when the spreadsheet changes. Delimiter and header row are detected, UTF-8 (with or without BOM) is expected.
• Text box templates reference columns through a csv mapping: {csv["Name"]}, or by position {csv[0]}. {row} is the 1-based row number.
• A current row with Previous / Next / First buttons and a preview of the row(s) the next job will use.
• Per copy: a workpiece shows row current + variable offset, so a grid array lays out one copy per row on the bed (batch on one sheet). A "copies needed" hint shows how many copies cover the remaining rows.
• Per job: with auto-advance on, the current row moves on by the number of copies after every finished job, so "one job per row" is just pressing Start again. When the last row has been used, the row stops there and a notification says so instead of wrapping.
• Copies whose row lies past the end of the file render the empty string for every column.

Scope I'd leave out for now: editing CSV contents inside Rayforge, other encodings than UTF-8, and a "one job per row" loop that runs unattended.

Describe alternatives you've considered
A --script template function can read a CSV file, but it cannot tell which copy it is rendering and it has no notion of "the row after the last job".

Additional context
Depends on #527 (it reuses the variable offsets and the auto-advance hook). I'm working on this and will report back with a branch.