[ENH] Add ability to re-use constructed C++ `graph_t` instances from Python

[rlratzel]

Current cuGraph Python APIs that require a call to an underlying C++ analytic will create a C++ graph_t instance prior to calling the C++ function, convert results to a Python/cuDF type, then delete the graph_t. This happens for each Python API call even if the intermediate graph_t is identical to the prior call. Constructing a graph_t can be expensive, and the time to construct it inflates algo run times.

note that graph_t instances are different than the pure-python Graph class instances, which are currently explicitly created and reused.

This issue tracks the work involved in allowing the Python layer to re-use graph_t instances if possible across algo calls. An additional API will also be provided to allow a user to construct the graph_t upfront in order to better track performance improvements/regressions of the underlying C++ code from python, since the time to construct a graph_t will then be excluded.

cc @seunghwak @ChuckHastings @BradReesWork

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions[bot]]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.