dials.two_theta_refine precision of p4p and cif output

[huwjenkins]

Running the small molecule tutorial data has flagged up small issues with the options in dials.two_theta_refine to output cell parameters (and their esds) in p4p and CIF format.

The result of

dials.two_theta_refine scaled.refl scaled.expt p4p=scaled.p4p cif=scaled.cif

is:

+-------------+-----------+----------------+
| Parameter   |     Value |   Estimated sd |
|-------------+-----------+----------------|
| a           |   5.42172 |        3e-05   |
| b           |   8.13331 |        5e-05   |
| c           |  12.0225  |        7e-05   |
| alpha       |  90       |        0       |
| beta        |  90       |        0       |
| gamma       |  90       |        0       |
| volume      | 530.15    |        0.00373 |
+-------------+-----------+----------------+
Saving refined experiments to refined_cell.expt
Saving CIF information to scaled.cif
Saving P4P info to scaled.p4p

but in scaled.p4p the esds for a and b have been lost:

CEL     5.4217 8.1333 12.0225 90.0000 90.0000 90.0000 530.1497
CELLSD   0.0000 0.0000 0.0001 0.0000 0.0000 0.0000 0.0037

and in scaled.cif something odd has happened to gamma although it is correctly listed with no esd:

_cell_length_a                    5.42172(3)
_cell_length_b                    8.13331(5)
_cell_length_c                    12.02248(7)
_cell_angle_alpha                 90.0
_cell_angle_beta                  90.0
_cell_angle_gamma                 89.99999999999999