1. Energy output per one piece of ore should be at least 50% of theoretical maximum.

A quad cell with 4 neighbors yields 140 eu/t, making that requirement translate to 17.5 eu/t per uranium cell (or a bit below 140 eu/t per ore). Assuming the quad cell lifetime bug gets fixed first, you can achieve that rate with a quad cell + single cell: 90 eu/t for 5 cells, or 18 eu/cell, for 172 heat/t.

Such a reactor can be made relatively cheaply, a Mark I in a 3-chamber reactor using no diamonds and 8 gold (4 overclocked vents). But that doesn't come close to your average output requirement.

3. Average output of reactor(s) should be at least 300 eu/t. If you breed in same reactor as you get energy from, breeding time is factored in when counting average energy. You can use more than one or two reactors, if needed.

300 eu/t puts us somewhere between a 2x2 square of 3 quads, 1 single (315 eu/t, 24 eu/cell, 744 heat) and a 2x2 of half quads, half dual (280 eu/t, 23 eu/cell, 640 heat). Or the simple single-cell version (the only one viable in 1.106 from an eu-per-ore perspective due to the quad/dual lifetime bug), a 4x4 square for 320 eu/t (20 eu/cell) with 656 heat. You might be able to manage something a little better with reflectors, but this seems like a reasonable rough number.

A 6-chamber reactor has 6*9 = 54 tiles. If we consume 16 of those with uranium, we have 38 left for heat management. At 656 heat, that means we need an average of 17.26 heat removal per tile. The only heat removal pieces close to that are component vent (16 heat) and overclocked vent (20 heat). Ignoring edge effects, they'd come out to 18 heat on average, so with edge effects they'll be well below our needed 17 heat. So, this reactor will need to be a Mark III or higher.