Science review - time horizons

Could we have further clarity on the time periods used? Jonathan Gregory's code had strict CF conventions used. Why do the projections start from 2006? e.g. the tas values should be time-means of calendar years 2006–2099, and zostoga should be values that apply at the year-ends, see below

This repository provides a [Python 3](https://docs.python.org/3) module called [ar5gmslr.py](https://github.com/JonathanGregory/ar5gmslr/blob/main/ar5gmslr.py.txt), containing a program to compute projections of annual-mean global-mean sea-level rise (GMSLR) for the ends of the calendar years 2006–2009 (or equivalently at the start of the years 2007–2100) with respect to the time-mean of 1996–2005, given annual global-mean projections of surface air temperature change (CMIP[56] variable tas) and thermosteric sea-level rise (CMIP[56] variable zostoga). The input should be supplied in netCDF files following the [CF convention](http://cfconventions.org/) with file names and contents as specified in the inline documentation of the project function. The tas values should be time-means of calendar years 2006–2099, and zostoga should be values that apply at the year-ends. The program checks that the time-coordinates and time-bounds are as expected, and aborts if they do not conform. For each quantity the program requires input timeseries of either

Start by reading the repository's projection entry points and compare their time handling with the referenced ar5gmslr.py.txt documentation and CF convention requirements. Check how tas and zostoga input times, bounds, and year ranges are interpreted; done means the expected 2006–2099 periods and year-end conventions are clarified and consistently handled.

Written by the indexing model from the issue text.