Improve new inflation estimation

[arhag]

From @zorba80's comment here:

When calculating new_tokens, we use the approximation 1 + r_c * dt for exp(r_c * dt) which is valid for small r_c and dt (dt approx 1/365). Now that we allow large inflation rates, this approximation might not be valid anymore (quadratic term might not negligible) and we have to use the full exponential instead.

[deckb]

With the introduction of staking based rewards and the move to vote based rewards to producers is this issue still relevant?