Normalize apte-hi dictionary headwords better, fix defects

[vvasuki]

Headword normalization

current headword normalization in apte-hi is adhoc (adding headwords from eliminating terminal म् and ः which does not follow dIrgha AkAra) and not that good. for example, we have a fake headword maruta as a result. This needs to be fixed.

to do this, someone needs to regenerate the dict using the below

• Initial data: JNU Apte S2H.txt.zip
• Headword normalization data - https://github.com/sanskrit-lexicon/hwnorm1/blob/master/sanhw1/hwnorm1c.txt

Other defects

The dictionary has other defects -

the uoh data had major errors in the entries i checked. - @akprasad

You can compare it with UoH's data (which mainly has the citations filled up) and the print/scan (which has more data that need to be inserted appropriately at resp. places throughout). - @Andhrabharati