reuse action clients for edges?

[gpdas]

In the current implementation, a client is created for each edge navigation corresponding to the edge action (edge_action_manager).

Is this or reusing action clients would be better / faster? Is it worth inspecting @adambinch ?

[Iranaphor]

Something that may be useful in this regard is instantiating the next edge client once the robot is some way through the edge. This kind of facility would show speedups across both inter-action routes and intra-action routes. However for how long it is between a robot reaching each new edge, and how long it takes to instantiate the edge action manager it does not seem necessary to rework the action client.

I will close this issue for now, but if we begin looking into more potential for further efficiency gains, this could be a consideration.