Difference in local time when calling physics in pure Matsuno mode

[emillour]

When running with "purmats=.true." (i.e. pure Matsuno time stepping) with identical iphysiq values, the universal time "ptime" sent to physics is different than when "purmats=.false" (default: i.e. Leapfrog+Matsuno time stepping).

Illustrative example: with day_step=960 and iphysiq=5

• if purmats=.false. then first call to physics is with ptime=0.0041666 (=4./960.) and the second one with ptime=0.009375 (=9./960.)
• if purmats=.true. then first call to physics is with ptime=0.005208333 (=5./960.) and the second one with ptime=0.0104167 (=10./960.)

So there seems to be a bad offset, but more likely in the "purmats=.false" case...