Hybrid Algorithm Refactor

[brrcrites]

The hybrid algorithms that were originally implemented were not implemented correctly, and so they have been refactored to be functional but essentially useless in their current form. Right now, they run either LMXRLF or DSATUR to find an initial coloring, and then use that coloring number as a condition to the Tabucol algorithm. The problem is that when that happens, Tabucol re-colors the entire graph to try and meet (and iteratively beat) the original coloring that was found. Instead, it should use the coloring that is generated from LMXRLF or DSATUR as an input, and then attempt to run tabu iterations on that initial graph rather than the random initialization that tabu would normally provide.

This will require a refactor of the Tabucol algorithm to have a separate initialization step, tabu step, and post-processing step. This way steps can be called when issuing a generic color() function call and a subset of steps can be called within other algorithms. We may want to make the tabu part public in case other people want to perform this process.

[brrcrites]

I'm proposing that we modify the Tabucol system into three main sub-functions, with a possible refactoring of the major tabu function. These would break up the color() function to include the following sub-functions:

• prep_graph(): this would take the this->graph that is assigned to the Tabucol object and reset any coloring that was previously assigned to it
• tabu(): this would run the tabu cycles to try and reduce the number of colors used in the graph
• make_min_contiguous(): this would contain the post-processing step to Tabucol introduced in Additional Test Cases & Edge Case Robustness #30 and the caller should be able to turn this off during construction if they don't want to re-color at the end

The tabu() step can then we utilized within the hybrid LMXRLF and DSATUR methods to actually try and reduce the generated initial coloring from those base algorithms, rather than its current situation which is to start a new coloring with the resulting number as the initial condition for Tabucol to meet.

[Sagebati]

I hope It's the good place for this remark.

In the readme we have "The algorithms hybridLMXRLF and hybridDSATUR are currently broken, see #29", If it's still the case why the cli defaults to hybridDSATUR, why the cli has the option to choose broken alogrithms.

[brrcrites]

@Sagebati this is a good point, I'll update the default method to use MCS since thats our suggested method for most people. The CLI has broken options essentially because they weren't always broken and I don't work on this repo all that often (I actually only saw this comment because I'm doing my yearly hacktoberfest upkeep)