cal2dec not robust

[gmaze]

The function cal2dec:
https://github.com/euroargodev/dm_floats/blob/c580b15202facaa0848ebe109103abe508d0dd5b/lib/+libargo/cal2dec.m
is probably not robust to leap years

This is fully discussed in euroargodev/argodmqc_owc#35

[gmaze]

This leads to the time stamps used in OWC to be wrong up to the minutes
This is not much, but still, it is wrong

Eg:
https://fleetmonitoring.euro-argo.eu/float/6903010

cycle 10 is reported at: 2021-03-02 05:38:30
but in a OWC create_float_source.m output, it is reported on 2021-03-02 05:37:59.999203200
this is off by 30 seconds

cycle 33 is reported at: 2021-10-28 05:28:30
but in a OWC create_float_source.m output, it is reported on 2021-10-28 05:27:59.999198399
this is off by 1 minutes

[gmaze]

Another eg:
https://fleetmonitoring.euro-argo.eu/float/6903075

cycle 1 is reported at: 12/12/2020 05:27:20
but in a OWC create_float_source.m output, it is reported at: 12/12/2020 04:08:59

cycle 2 is reported at: 22/12/2020 05:52:20
but in a OWC create_float_source.m output, it is reported at: 22/12/2020 05:13:30

These are more than 30 minutes apart !