Handling leap years

[tomjemmett]

Currently, we take a baseline year and project it forwards.

We do not currently handle leap years. Using a baseline of 2023/24 would include activity for 29th Feb 2024. If we project forward to a non-leap year then we will be including 1 extra day of activity.

Likewise, if we start with a baseline year that isn't a leap year but project to a leap year we will be short 1 day of activity.

There are a couple of approaches to this that spring to mind:

0. ignore it as a small issue that wouldn't have enough of an impact to worry about
1. we could add a correction factor based on the start/end years to sampling by $1\pm \tfrac{1}{365}$
2. we could assume a year of 365 days by dropping 29th Feb and make this a clear assumption (e.g. with capacity conversion to always have a year with 365 days)