mv_machine_learning: drop template from MachineLearningConfig class

mv_machine_learning: drop template from MachineLearningConfig class

[Issue type] code refactoring

MachineLearningConfig class introduced in [1] required the
ParserType template to initialize the _parser. This resulted in
the inclusion of parser header files of each task group in the
higher level MachineLearningConfig class which led to code smell.

This patch drops the use of template and as a first step,
modifies the object detection 3d task group to work with this
new version.

[1] https://review.tizen.org/gerrit/#/c/platform/core/api/mediavision/+/301043/

Change-Id: I90971294c89615d85ae61e789fa6200e53d367c9
Signed-off-by: Vibhav Aggarwal <v.aggarwal@samsung.com>