Adjusts the time stretch calculation to hold the ratio of reserves constant

[jalextowle]

Fixes: #715.

The gist of this PR is that all we need to do to maintain an approximately constant inventory across pool's with similar initial conditions but different position durations is to solve for a time stretch that implies the same ratio of reserves.

To derive this, we use the spot price calculation:

$$ p = \left( \tfrac{\mu \cdot z_e}{y} \right)^{t_s} $$

We start by solving for the benchmark time stretch $t_{s_b}$ using the existing calculateTimeStretch function. Then we use this time stretch to solve for the ratio of reserves $ratio = \tfrac{\mu \cdot z}{y}$ using calculateInitialSharePrice. For simplicity we use a benchmark of $z_e = 1$ and $\mu = 1$, but this won't effect the ending result since the ratio will be the same regardless of the starting initial share price and effective share reserves. Finally, we solve for the target spot price $p_{target}$ from the position duration $d_{position}$ and target APR $r_{target}$:

$$ p_{target} = \frac{1}{1 + r_{target} \cdot \tfrac{d_{position}}{365\text{ days}}} $$

Finally, we use the spot price calculation to solve for the adjusted time stretch:

$$ p_{target} = \left( \frac{\mu \cdot z_e}{y} \right)^{t_s} \implies t_s = \frac{\ln{p_{target}}}{\ln{\tfrac{\mu \cdot z_e}{y}}} $$

NOTE: Any logarithm will do, but we use the natural logarithm since we have an efficient fixed point implementation in Solidity.

[coveralls]

coverage: 95.057%. remained the same
when pulling ff99411 on jalextowle/fix/time-stretch-adjustment
into 62cec39 on main.