Thanks for the feedback!
I fixed the coding style and RMB tooltips. For this one I had to change a bit the mouse handling logic of window.cpp, I'm not entirely sure what I did is valid so I put it in a separate commit.

On the performance of LinkGraphOverlay::ShowTooltip, its complexity is linear in the number of links for the selected cargo type.
I agree this is a bit much for large networks, so now hover events are only generated every _settings_client.gui.hover_delay_ms when the cursor is static (they were generated at each frame before). With the default setting of 250ms, ShowTooltip will be called at most 4 times per second.

Might we get a Preview label to make testing this easier?

Examples of the tooltip:

It appears to only work when you have exactly one cargo type selected in the cargo flow legend, if I select more than one cargo type the tooltips won't show up.

It appears to only work when you have exactly one cargo type selected in the cargo flow legend, if I select more than one cargo type the tooltips won't show up.

I just pushed a new version that displays statistics for the most saturated one when a link has several cargo types. This is already how the color to use for a link is determined, so the two will always match.

In my opinion, this and #9866 close #9167 and #9168.

One point of minor confusion for me regarding links which are over capacity — e.g. 219 passengers transported back (547% of capacity) in 10 days.

To me, this reads like 219 passengers is 547% of capacity. Obviously it's not, but perhaps there's a better way to word this. What about 219 passengers transported back in 10 days (547% of capacity)?

219 passengers/month is 547% of capacity. Even without this PR, all dark red links in the overlay correspond to links with a 200%+ usage/capacity ratio.

The "capacity" of a link is only a rough estimate, based on the capacity of the vehicles that took the link in the past few months.
As for the usage of a link, it is also approximative, and it's a bit more complex as the value used is actually max(planned, usage), with 'planned' an (even rougher) estimate of the future demand based on the number of passengers waiting for this link.

By construction, the estimated usage of a link never exceeds 100% of its capacity, but the demand can, thus the occasional >100% usage ratios. This is suboptimal, but reworking the formulae to be more accurate is definitely out of scope for this PR. I may give it a try at some point in the future though :)

In any case, I realize the tooltip text is a bit confusing as is.
I read your proposed wording as '219 passengers crossed this link the last 10 days' when the intended meaning of the tooltip is actually 'about 219 passengers cross this link each month, and the journey takes them 10 days on average'.
Maybe 219 passengers transported back per month (547% of capacity) in an average of 10 days would be less ambiguous? I'm open to alternative wording suggestions!

it looks like there are two different stats trying to be condensed into a single line here.
how does this sound?
219 passengers (547% of current capacity) to be transported back per month
average return journey time: 10 days

that stops it from sounding like more than 100% capacity is being transported, and also helps prevent capacity and time from being confused.

Thanks for the suggestions. I slightly reworded the tooltip:

X passengers to be transported per month from A to B (x% of capacity)
Y passengers to be transported back (y% of capacity)
Average travel time: n days

The travel time is now on a separate line, and is an average for the two directions of travel.

The to be transported helps a lot, indicating demand versus actual stats.