[DOC] Obtain algorithm performance trends on large data sets

Currently performance compares cuGraph performance vs NetworkX. It does not allow performance measuring vs older cuGraph versions.

• Add an ability to provide old run data and compare it to running new data. Suggest this be done by storing the cuGraph results from older runs and include them in the output data.
• Run the notebook vs old release containers (22.04, 22.06. 22.08)

For example, store the following from a previous run and check in the notebook and do a comparison of run times.
print (time_algo_cu)

[[0.17079168488271534, 0.3395446129143238, 0.09437482920475304, 0.03897589887492359, 0.06541383406147361, 0.07896647695451975, 0.03612494096159935, 0.03178467997349799, 0.050747953122481704, 0.05190548300743103], [0.06335067399777472, 0.4162313530687243, 0.2838626359589398, 0.07679838291369379, 0.10212528496049345, 0.15321092098020017, 0.048983687069267035, 0.034497485030442476, 0.05879357596859336, 0.06953499908559024], [0.4182498489972204, 1.2072639458347112, 0.6116538709029555, 0.4192121799569577, 0.34089540387503803, 0.9776100679300725, 0.2997500051278621, 0.2437728471122682, 0.7524426199961454, 0.4012443018145859], [0.32816758984699845, 3.907691675936803, 0.9266399221960455, 1.101785965031013, 0.3137800102122128, 1.2585107909981161, 0.3471053291577846, 0.18662777706049383, 2.1188537278212607, 0.34518214990384877]]

Since release notebook will be reworked soon. This issue should be updated to reflect the new notebook.