Request to add certain training parameters present in classic Lightgbm to Synapse ML

[alzio2607]

Is your feature request related to a problem? Please describe.
Some features present in the classic (non-spark) Lightgbm are not there in synapseML. These features are really needed for someone looking to switch from the classic way to Synapse . It'll be appreciated if the parameters are integrated since they help with more precise training too.

Describe the solution you'd like
Would like to request the addition of the following params
cat_l2
max_cat_threshold
cat_smooth
min_data_per_group
min_data_in_bin

Additional context
Found 2 open issues pertaining to similar kind of request. Mentioning them here too. Creating this new issue for latest update
#997
#789

AB#1799267

[imatiach-msft]

@alzio2607 I think this is somewhat resolved with PR:
#1444
which was merged recently, since

Add ability for LightBGM user to set custom args that aren't defined explicitly by SynapseML (as Vowpal Wabbit does now)

adding @svotaw
Ofcourse, we should add these parameters to the API directly as well, in addition to the custom args parameter.

[svotaw]

You should be able to set those as well as any other manual parameter after the above changes.