Should roocs provide "warnings" as well as "fixes"?

[agstephens]

At present we use the "character" to provide "fixes".

In some cases, there will be issues about certain data sets that cannot be fixed - but will be useful for users to know about.

Examples:

1. A model that has not provided the same time range as others for a given experiment. There is no fix, but if users know it they can modify their approach to exclude the model or modify the time range they are using.

2. All data for an experiment uses a virtual time range (e.g. control runs). In such cases the actual time is irrelevant - so different models start at different years. This could be provided in "warnings" - so the user can approach the data accordingly.