Figured bass issues

• input/regression/figured-bass.ly: Clean-ups

• input/regression/figured-bass-*.ly: Format, whitespace, simplify, normalize

• format-bass-figure: Minor code shuffling

• Improve rendering of BassFigureContinuation

  Up to now, the code simply used markup string "0" as an invisible markup text for the right side of the BassFigureContinuation spanner to get its correct length. The new code uses digit 0 (which corresponds to input \figuremode { <0> }) and sends it through figuredBassFormatter.

  As a consequence, the figured bass extender line really ends at the right edge of a figured bass stack by default, and BassFigureLine grobs now have the same height with and without extender lines – the previously used "0" markup string resulted in an ordinary digit from a text font, which is usually much larger than a figured bass digit from the Emmentaler font and thus vertically enlarged the whole BassFigureLine grob.

• Make horizontal-line-spanner-interface stand-alone

• Use horizontal-line-spanner-interface for figured bass continuation lines

  The previously used padding property was not sufficient to cater for necessary line length adjustments.

• Figured_bass_engraver::process_music: Fix event handling

  If the current moment is within the time span of the last figured bass event, don't clear spanners. Without this correction it is possible to get an incorrect second BassFigureLine grob, eventually positioning the invisible bass figure (which is used to find the continuation line's right edge) too high.

  We already have a regression test for that (figured-bass-durations.ly) but this detail was apparently missed – the placement of the invisible bass figure can only be seen with activated skyline debugging.

  Fixes issues #1513 (closed) and #6428 (closed).

• Correctly handle figured bass continuation line after empty bass figure

  Use an invisible figured bass digit to start a continuation line; this triggers correct horizontal and vertical positioning.

  Fixes issue #6540 (closed).