Plotting functionality should join `historical_activity` to `ACT`

[glatterf42]

@glatterf42 and @khaeru, I needed to filter the years explicitly here, because the plotting module does not show the values in historical years, which in this case is 690. Ideally, the plot should join historical_activity (for year 690) to ACT (for years 700, 710, and 720). If not possible, the plotting should not print historical year (690) without showing any value for that, which is the default behavior now.

_Originally posted by @behnam-zakeri in https://github.com/iiasa/message_ix/pull/815#discussion_r1553163224_

This was said with respect to this line.

[khaeru]

I think we could actually make several such enhancements. For example, the default reporting config includes: https://github.com/iiasa/message_ix/blob/7dce4617b2ca6ca17a03cbe20fcc8e57fd56fbae/message_ix/report/__init__.py#L64 …but there is also a parameter: https://github.com/iiasa/message_ix/blob/7dce4617b2ca6ca17a03cbe20fcc8e57fd56fbae/message_ix/models.py#L433

We can try to do concatenation wherever there are such pairs.

I say ‘try’ because some issues may come up:

• In some cases the dimensionality of the historical_* parameter is mismatched with the product of 1+ variables and 0+ parameters to which it supposedly corresponds. This is true in the case above.
• message_ix_models.report.legacy has a parameter mergehist, which triggers concatenation of IAMC-structured time series data for pre-model periods with computed values for model periods. This points us to the possibility that a user populates both these historical time series data and the `historical` parameters with potentially different* values. What should the reporting do in such cases?