OK, the sketch helps. I was not visualizing it properly.
I see the problem and agree that a flat bottom end mill would be ideal.
But, you could still do it with a standard, slightly concave end mill. Referring to the sketch, start with the end mill's axis directly over the rotational axis of the work and make a full rotation of the work. That would cut the corners of the slot to the full depth and leave them square. Then offset the end mill by some small increment, tangent to that rotational arrow and make another complete rotation of the work. That should cut two paths just inside of the corners to full depth. Continue moving the end mill in small increments and rotating the work for each until the edge of the end mill is directly over the rotational axis of the work. That should produce a real close approximation of a flat bottom for the slot.
If you just want to to eliminate that center hump, then two positions of the end mill, one radius apart, should do it.
Rough calculation:
You specified a 3/4" end mill so it is 3/8" (0.375") from center to the corners.
If the cutting edges have a 1 degree angle, the height of the center hump is 0.375" TAN(1) = 0.0065"
Assuming a 0.001" hill in the surface is acceptable (+/-0.0005").
0.0065" / 0.001" = 6.5 steps.
So stepping the end mill over 7 times by a 0.536" will result in a surface flatness of less than +/- 0.0005".
The trade-off here is more steps produce a flatter surface and fewer steps a rougher one.
And, yes the machining time would be increased. Factor that time against the cost of a special end mill. It may depend of the total number of parts needed.
No, as I understand it this is incorrect and irrelevant for what the O.P. is doing. He is using a rotary or 4th axis to rotate the work past the cutter, not moving the cutter in a straight line past the work. When this is done and the work is rotated past the end of a milling cutter, a straight line is not produced unless the cutter has a flat bottom. This is the operation we're talking about - milling cutter on center and work being rotated:
View attachment 249263