The bulk (substrate) connection are shorted, so it is clear that you can put the transistors in the same N-well.
Of course, you can place the transistors in separate N-wells, and connect by metal the two n-wells. There is no electrical advantage for this approach. In addition, the disadvantage is the area increase. Layout design rules ask, in general for quite big distance between two separate n-wells. This comes from technology - n-well is a deep implant, with long annealing time, therefore big diffusion after the implant.
The second disadvantage is mismatch. Placing the transistors in separate n-wells ould result in a big distance between them, therefore bad matching.
Have fun!
Added after 2 minutes:
Reading more carefully your question, I have a question:
Why do you want to build a current mirror with two different transistor type?
Is better to build the mirror using only one type (better matching), in you case, probably because of voltage constraints, 2.5V type.