Copasul

* new glob features stored in `copa.data.fi.ci.glob.si.eou` as
* `.{tl|ml|bl|rng}_drop`: actual f0 drop in IP, i.e. dur * rate
* `.tl_ml_cross_t|f0`, `.tl_bl_cross_t|f0`, `.ml_bl_cross_t|f0`: time and f0 of crossing point of respective line pair
* content is in glob table columns `{tl|ml|bl|rng}_drop` and `tl_ml_cross_t|f0`
* **Initial commit on Gitub**

* new options `.fsys.augment.{syl|loc}.force = <0>|1`. Dependencies: if `.fsys.export.summary==1`, both `.force` are set to 1. If `loc.force==1`, also `syl.force` is set to 1. If `syl.force` is 1, at least one syllable nucleus is to be found per parent tier segment. In case no nucleus is extracted within a segment, a nucleus is placed at the energy maximum. If `loc.force` is 1, at least one pitch accent is to be found in a file. In case no accent is extracted, an accent is placed at the energy maximum. Both ensures, that for all files (also e.g. very short or noisy ones) all selected feature sets can be extracted, which is needed for `.fsys.export.summary==1`, so that all columns have the same number of rows.
* `copasul_augment.aug_loc_fv()`: bugfix; adjusting number of return arguments

* new loc features: `{bl|ml|tl|rng}_d`; mean distance between local regression line and corrsponding part of global regression line. The higher, the more the local is above the global line.

* `copasul_augment.aug_batch_ps()`: bugfix; adjusting number of return arguments of `aug_glob_fv()`

* `copasul_augment.<divers>()`: increasing robustness.

* `copasul_styl.styl_discont()`: for each register representation 2 more features of use e.g. for measuring downstep, namely `d_o`: onset difference, and `d_m`: difference of means