Handle unmatched strategies (`-`) in independent change-factor charts

[matt-dray]

You may have explained this in the past, @tomjemmett, or I'm missing something obvious: I don't remember why we allow - in the independent change-factor charts.

Inspired by this observation: The-Strategy-Unit/nhp_scenario_comparisons#160 (comment)

Some code to replicate: The-Strategy-Unit/nhp_scenario_comparisons#160 (comment)

Two things:

1. convert_to_tele doesn't get matched to a strategy, so shows up as -.
2. change_factors that aren't strategies don't get matched, of course, so also end up as -.

Re (1), maybe we should ifelse() to insert "Convert to tele" into the strategy column. Re (2), maybe these should be filtered out if they add nothing to the chart? Or is the idea that we want to show all contributing factors? If so, perhaps we should be providing strategy names even if they aren't strategies.

[francisbarton]

This is what reskit is currently planning to do around this - happy to correct if that filter is not the right logic.

[ghobro]

The sister issue here is The-Strategy-Unit/nhp_scenario_comparisons#142 as The-Strategy-Unit/nhp_scenario_comparisons#160 is more about the fact that certain TPMAs are causing more activity in an activity type - measure combination due to them being about shifting care