For now, I have no plans to revise the atan2 code. If somebody else
wants to, that's great. For what it's worth, my thinking now is that
I would:
atan2(x,x) --> atan2(x,x) (simplifies to itself)
assume(x > 0);
atan2(x,x) --> %pi / 4
After all, abs(x) doesn't simplify to a conditional (or to my favorite
max(-x,x)). Well maybe abs should simplify to a conditional expression,
but I'd guess that few would think that the conditional expression
is a *simplification*.
There is overlap between atan2, carg, polarform, and ... Any revision
for atan2 might be able to blend these functions.
Barton