updated NTT block to support masking countermeasure for A.y

[mojtaba-bisheh]

The current masking supports only PWM operation that is followed by INTT operations. Hence, the intermediate results of PWM is not stored in memory.
In the case of A.y, PWM with accumulation needs to happen first and then INTT. Since this is accumulation over several polynomials, we cannot perform INTT on the fly for each PWM operation that happens.
We should update the memory arch to store the intermediate masked data of size (4 coeff * 2 shares * 46 bits * 64 addr) that stores 1 poly at a time and accumulation happens over it with masked PWM. Then INTT is triggered that reads this share memory and performs 1st stage of masking and combines shares and writes to original memory.

Now, NTT should support:
All unmasked operations
Masked PWM
Masked PWMA (accumulation)
Masked INTT
Masked PWM + INTT