Let us accelerate craft in a straight line. Let rate of aceleration be
A.

Let us now put a kink in the course. Let velocity be v craft having
accelerated for v/A seconds. Let us have a small acceleration a to the
side.

Total gain in velocity will be A-a^2/2 assuming aA

After a further t seconds velocity will be v+(A-a^2/2)t,at or
(v+A)t-t(a^2-a^2v/(At+V)

This is always less than v+At the velocity it would have if it had been
accelerated in a straight line. The distance travelled will be the
integral of this.

In fact small kinks reduce the distance to the accelerator but only by
reducing total velocity. Hence the optimal is a straight line.
Straightening out the king is always an improvement. One never improves
the situation by accentuating a kink.