The CMake process will pick up Qt's Qt5Core_VERSION_STRING and
stick it in a cache variable (with forced updating each run),
and a new `#define` in the header exports it to the library and
to the Python bindings so that it can be queried at runtime.
- Rename to match package naming used elsewhere (official configs)
- Add IMPORTED target creation to the Find module, along with
version number detection
- Switch to lowercase variable forms (e.g. jsoncpp_FOUND)
The backward conversion int16_t to float in libopenshot has range
(1.0; -1.0], thus conversion -1.0f to int16_t should be secured.
Float values can exceed the (1.0; -1.0) range. This can cause
distortion in audio instead of limiting values at max/min for the
int16_t.
* bindings: Remove relative header paths
Use the include paths set on the command line the way
they're supposed to be used.
* Also remove some redundant include paths
* Fix SWIG include dirs
Incorrect conversion of float to int16_t caused distortion that was
mistakenly perceived as clipping. Now because the convertion was fixed,
there is no more sense to reduce input gain of the source.
