Feature/issue 150 duration unit

[arshiacont]

Fixes #150

Add a dur unit possibility to TagParameter and parsing in quarter-note.
This feature is fully backward compatible and should not break current behavior.

Info:

• Adjusted duration-based tag positioning to look up the measure that actually contains the target time (can cross barlines) and compute offsets from that spacing instead of clamping to the current measure; keeps the debug trace for visibility. src/engine/graphic/GRTag.cpp
• Made measure-bound lookup robust even when only the upcoming barline is present by scanning all barlines, choosing nearest previous/next, and deriving a previous start from the meter (with a 1/1 fallback). This feeds correct widths to duration-based offsets. src/engine/graphic/GRStaff.cpp
• Apply duration offsets only after position tags finish anchoring so tags have stable positions before we shift them. src/engine/graphic/GRStaff.cpp

[dfober]

Sorry for the slow response. I had started working on it (with a new unit named 'rl' - for relative location), but you were faster.
I have several questions:

• I didn't quite understand what duration the new unit 'dur' refers to: is it a whole note, a quarter note?
• looking at the source code, the strategy adopted is based on bars, which means that without bar it shouldn't work. For my part, I was thinking of evaluating the space between 2 successive signs (which frame the tag) to calculate the displacement, which is a bit more complex but more general.
• I compiled the code but I couldn't get the new unit to work

[arshiacont]

Test and see it on regression-tests/g-issue150.gmn :

• That's why I used bars: in that example you have several harmonies on a single note with nothing else! Using bars (and my limited knowledge of Guido Internals) was the only way. :)
• "dur" is in whole note but in Double (I started with fractions but abandonned before getting the parser too messy).
• I made it work on regression-tests/g-issue150.gmn meaning on Harmony tags! Didn't go much far as to test it on other units.

We can keep iterating this PR until we are satisfied. I will try to setup more unit tests.

[dfober]

OK. I managed to get it working but it doesn't work as expected, e.g. with this simple case:

[
	\meter<"2/4">
	\i<"p", dx=0.25dur> a  b c a d f g f e d c
]

the 'p' is shifted to the 'c' but it works when the \i is placed after the first 'a'
Maybe using the mapping (already computed) could help for a more general solution. I'll give it a try (but I'm a little overwhelmed).