Hard blocker — drill.c case 'R' → case 'S'

After processing a repeat-hole R command in the body section, the code falls through into case 'S' (spindle speed), which logs a spurious note and calls eat_line(fd) — consuming the next line from the drill file. That next line is typically a coordinate, tool change, or M30 end-of-file command, which gets silently discarded.

case 'R':
    if (state->curr_section == DRILL_HEADER) {
        // log error
    } else {
        // parse repeat count and step values
        // emit rcnt drill holes
    }
    // [[fallthrough]]; <-- PR annotates this as intentional
case 'S':
    gerbv_stats_printf(..., _("Ignoring setting spindle speed..."));
    eat_line(fd);
    break;

The existing golden-image test passes because it was generated against the same buggy code. The correct fix is a break; before case 'S':, not a [[fallthrough]];.

Likely bug — callbacks.c RS274XM → DRILL

On the success path for CALLBACKS_SAVE_FILE_RS274XM, the code falls through into the CALLBACKS_SAVE_FILE_DRILL block which overwrites windowTitle (leaking the RS-274X string) with a drill-file dialog title and sets the file extension to .drl. An RS-274X merge export should not be presenting a drill-file dialog. This looks like a missing break; that has always been there, with the PR locking it in as intentional.

Correct annotations (6 of 8)

The three existing /* fall through */ comments in scheme.c are correctly left untouched — GCC level 3 already accepts them.

When you fixed it, amend it into current commit and push it --force--with-lease.

Fixed both — drill.c case 'R' and callbacks.c RS274XM now have break; instead of [[fallthrough]];. Force-pushed the amended commit.