Tibetan and SEA transliteration errors (ALA-LC)

[scossu]

We have run some tests with Aksharamukha integrated in an external app. Tibetan and other South East Asian strings transliterated to/from ALA-LC Roman are displaying some issues. Attached is a battery of test strings that failed.

I spot-checked these strings in the Aksharamukha web interface and I can confirm the discrepancies from the expected results.

sea.csv
tibetan.csv

Note: the first column is a language code used in our app, but it should be easy to guess the language used in Aksharamukha.

Thanks for looking into this.