Refactor storage index

[cangfengzhs]

What type of PR is this?

• bug
• feature

Which issue(s) this PR fixes:

Is part of issue 2745

What this PR does / why we need it?

In this PR, the index execution plan in storage is completely rewritten. In the original Index execution plan, querying index data and querying base data are completely separated, and the kv pairs of base data or kv pairs of index data are passed between the nodes of the execution plan, which leads to serious coupling between nodes , And can not deal with the problem of String type index being truncated.

This PR logically mainly contains the following modifications

1. Graphd no longer performs interval conversion when the expression is converted to ColumnHint, and forcibly converts all intervals into left-closed and right-opened intervals, but transmits the original interval opening and closing information to storage without any truncation, otherwise the query will be found The result is wrong. (Currently only storage has been modified to adapt to this logic, but the modification of graphd requires another PR)
2. Each node in the execution plan of the Index transfers Row instead of the original key-value uniformly. In addition, the processing method has been completely changed to iterative processing (may have a slight decrease in performance, not very sure. If necessary, batch processing can be added).
3. Added a unittest to the Index execution plan. The previous unittest can no longer be compiled successfully.
4. Encoding of double type index. Just a draft, in order to pass the unittest.

Special notes for your reviewer, ex. impact of this fix, etc:

This is a PR that has not yet been fully completed, and some comments need to be added, as well as the processing of some corner cases, and the supplement of unittest.

Just look at the overall execution logic.

Additional context:

Checklist：

• Documentation affected （If need to modify document, please label it.)
• Incompatible （If it is incompatile, please describle it and label it.）
• Need to cherry pick （If need to cherry pick to some branchs, please label the destination version(s).)
• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory

Release notes：

Please confirm whether to reflect in release notes and how to describe:

                                                            `

[critical27]

Generally LGTM

[pengweisong]

will this be removed ?

[bright-starry-sky]

LGTM

[critical27]

I'm ok with this PR

[critical27]

Well done, great work

[codecov-commenter]

Codecov Report

Merging #3196 (2ff0617) into master (58f0b44) will increase coverage by 0.08%.
The diff coverage is 91.19%.

@@            Coverage Diff             @@
##           master    #3196      +/-   ##
==========================================
+ Coverage   85.24%   85.32%   +0.08%     
==========================================
  Files        1295     1304       +9     
  Lines      118190   119900    +1710     
==========================================
+ Hits       100748   102306    +1558     
- Misses      17442    17594     +152     

Impacted Files	Coverage Δ
src/storage/ExprVisitorBase.h	0.00% <0.00%> (ø)
src/storage/exec/StorageIterator.h	94.33% <ø> (+5.45%)	⬆️
src/storage/test/LookupIndexTest.cpp	100.00% <ø> (ø)
src/storage/exec/IndexSelectionNode.h	26.47% <26.47%> (ø)
src/storage/ExprVisitorBase.cpp	29.31% <29.31%> (ø)
src/storage/test/IndexTestUtil.h	66.53% <66.53%> (ø)
src/storage/exec/IndexLimitNode.cpp	66.66% <66.66%> (ø)
src/storage/exec/IndexVertexScanNode.cpp	90.41% <90.41%> (ø)
src/storage/exec/IndexNode.cpp	91.66% <91.66%> (ø)
src/storage/exec/IndexScanNode.cpp	92.50% <92.50%> (ø)
... and 52 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 58f0b44...2ff0617. Read the comment docs.

[liuyu85cn]

LGTM