Updates PR #627 - Suffix Arrays

[stress-tess]

This pull request updates PR #627 submitted by @zhihuidu.

This PR:

• Brings string-suffix-array-functionality up to date with master
• Rearranges code to increase separation between strings and suffix arrays and minimize diff with master/highlight changes
• Removes external libdivsufsort in favor of zhihuidu's native chapel suffix array construction implementation
• Changes formatting in some areas

---

bf1757c Repackages zhihuidu's commits in PR#627

• Merge squashed commits and rebased over a fresh copy of master. This makes the commits more portable and the git history of master more readable with all commits grouped together
• Helps minimize diff with master since I default to master in conflicts where there are only formatting differences

07f51d8 Applies original minimum changes to resolve conflicts with master

• Updates logging statements to use logLevel instead of v
  • PR replaced --v flag with LogLevel enum config and implemented module-level logging #728 #757 #760
• Updates check to check_table
  • PR Closes #791: Remove duplicated check table in MultiTypeSymbolTable #792
• Removes outdated Strings.attach function which used _ for subcomponents (new convention is . )
  • PR Issue #761: Categorical, Strings, and pdarray registration updates #774
• Makes unregister_strings_by_name a staticmethod
• Applies recent changes to SegmentedArray.chpl (automatic merge was thrown off by addition of class SegSArray) including:
  • PR 646 string test test ends with periodically fails #666
  • PR Closes #762: Fixes off by one error in Segemented array peel method. #781

ca4941b Implements changes suggested in PR 627 feedback

• Minimizes diff with master to highlight functionality changes
• Moves SArrarys class to suffix_array.py
• Refomats suffix_array.py
• Reformats SACA.chpl to fix bracket alignment
• Removes commented out code
• Adds type hints to size and bytes in strings.py for mypy
• Replaces writelns in SegmentedArray.chpl with logging statements following the example of SegString class

Previously stand-alone commits have been squashed into this one since Zhihui approved them

• Removes external libdivsufsort code and references. Defaults to zhihuidu's native Chapel suffix array construction implementation using skew algorithm (based on Simple Linear Work Suffix Array Construction)
• Using the native chapel suffix array construction implementation prevents copying the libdivsufsort codebase into arkouda or forcing the dependency on arkouda users.
• Renames class SegSArray to class SegSuffixArray and moves this class to SegmentedSuffixArray.chpl (previously in SegmentedArray.chpl along with SegString). Rearranged to increase separation with strings and minimize diff with master.

935f29b Implements changes suggested in this PR's feedback

• Removes unused fields variable in _parse_single_int_array_value
• Passes Exception into ValueError in _parse_single_int_array_value
• Inverts logic of in1d_int to avoid duplicating the ind1d logic
• Adds self.name to SArray class and updates attach/register functions to use dots instead of dashs to be consistent with the Strings class
• Removes unnecessary cast in suffix_array_file
• Changes typing from sipHash128 and computeSipHashLocalized to accept int and uint(8)
• Updates suffix_array registration and info functions
• Renames SACA.chpl to SuffixArrayConstruction.chpl
• Adds UnitTestSuffixArrayConstruction.chpl
• Adds docstrings to SuffixArrayConstruction.chpl and removes duplicated logic by modifying radixPass to accept int or uint(8)

[stress-tess]

@zhihuidu, are you okay with the changes I've made here? I specifically made the move/rename class SegSArray and remove libdivsufsort commits stand alone, so they are easy to revert

Once I have your approval, I'll invite the other core team members to review

Edit: I've heard back from zhihui and he is happy with this restructuring

[stress-tess]

@zhihuidu, I rebased over master so the suffix_array branch is up to date with no merge conflicts. I also made some slight changes to SuffixArrayConstruction to shorten the code. I verified the functionality remained the same with python3 benchmarks/sa.py localhost 5555 --correctness-only

[zhihuidu]

@zhihuidu, I rebased over master so the suffix_array branch is up to date with no merge conflicts. I also made some slight changes to SuffixArrayConstruction to shorten the code. I verified the functionality remained the same with python3 benchmarks/sa.py localhost 5555 --correctness-only

@pierce314159 I appreciate it. In fact, you almost rewrite the code.

[mhmerrill]

@pierce314159 @zhihuidu I would like to see this updated and moved to the arkouda-njit or arkouda-contrib so it can be included as an optional module in the arkouda build.

[zhihuidu]

@mhmerrill @pierce314159 No problem. I can rewrite the suffix array code based on Kyle's framework to integrate new functionalities and push it to arkouda-njit first.

[stress-tess]

@mhmerrill @pierce314159 No problem. I can rewrite the suffix array code based on Kyle's framework to integrate new functionalities and push it to arkouda-njit first.

@zhihuidu It might be easier than that, let me take a crack at it and see if I can use git magic to get it in line with the njit repo

[zhihuidu]

@pierce314159 If some tools can do it automatically, it will be better. Please go ahead.

[glitch]

@mhmerrill @pierce314159 No problem. I can rewrite the suffix array code based on Kyle's framework to integrate new functionalities and push it to arkouda-njit first.

@zhihuidu It might be easier than that, let me take a crack at it and see if I can use git magic to get it in line with the njit repo

Note: It looked like @pierce314159 didn't have access to the repo, that should be fixed now. 👍

[mhmerrill]

I am going to close this because we moved this to Bears-R-Us/arkouda-njit#3