Possible origin of arsha variations

[Abhishek Mehta]

I was thinking about the usage of arsha variations of the panini dhatupatha, for instance त्यजँ is a परस्मैपदी भवादिगणः धातु but for instance in the Valmiki Ramayana one would find an आत्मनिपदी variation used. For instance here 

I think the above usage can be justified if the original dhatu was त्यजञ् instead of त्यजँ so that स्वरितञितः कर्त्रभिप्राये क्रियाफले sutra kick in and force an आत्मनीपदी usage. 

I have a vague memory of an online tutorial I watched where I heard that dhatus like त्यज has an assumed अजनुनिनासिक इत्. If this is true then the परस्मैपदी/आत्मनिपदी usage of त्यज is dependent on what the author assumes to be the इत् that is happening. If anyone understands what my vauge memory seems to be referring to please let me know the valid references that back it or correct it if it's false. Also is there an alternative to panini dhatupatha that can also explain this?