Small fixes to InferenceEndpointModel (#112)
- Added implementation for missing properties in InferenceEndpointModel
- Added tokenized context in greedy generate + handled stop_sequence of tuple/list/str
- Santized endpoint model name (to extend later)
- Redid disable_tqdm & fixed call to batch generate with logits
- Removed debug flag & swapped debug to true
- Added get original order for InferenceEndpoint calls
