[LPT] Remove mandatory weight folding in ConvolutionTransformation (#35969)
### Details:
- Removes weight dequantization in `!canConvolutionBeTransformed()`
cases
- Serves to prevent improper weight constant folding on false matches
from activation multiplication
### Tickets:
- CVS-186993
### AI Assistance:
- AI assistance used: no