Use square DP for integration when narrow resonances found in m12

[Thomas Latham]

As mentioned in comment:1:ticket:48, having narrow resonances on the diagonal (i.e. in m12) can severely slow down the integration. The reason for this is that, at present, we just use a very fine grid over the whole DP in such a case. A better solution could be to integrate in the square DP instead.

[Thomas Latham]

(In [308]) Switch to integrating in square DP when narrow resonances are found in m12. Basic implementation is there, so oktotest #65. Once verified, need to add handles (or reuse existing ones) to tune grid size etc.

[Thomas Latham]

(In [310]) Add ability to set the integration grid size when integrating in the square DP, addresses #65. Still need to consider the automatic division of the sqDP around narrow resonances in m12 (and hence mPrime).

[Thomas Latham]

Since the default grid seems to give a good resolution for most resonances it seems unnecessary to complicate things further. If the need arises this can be added in future.