Consolidate implementation of _process_totals

[Swatinem]

In particular, this unifies the duplicated implementations of these methods across Filtered/ReportFile.

The new method also spells out a "very concise, but also unreadable" implementation that sums up the cyclomatic complexity (something I would like to remove completely eventually). This is now more readable and understandable, but more importantly, it only iterates over the lines once, instead of twice as the previous implementation did. Along with avoiding a bunch of intermediate list allocations that the FilteredReportFile implementation did, this should yield quite a good speedup.

---

I get the following benchmarks results now locally (as sadly, benchmarks are not yet fully working in CI):

                                         Benchmark Results                                          
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━┓
┃                                       Benchmark ┃   Time (best) ┃ Rel. StdDev ┃ Run time ┃ Iters ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━┩
│                     test_report_parsing[Report] │       7,184ns │        2.9% │    3.02s │ 5,396 │
│             test_report_parsing[ReadOnlyReport] │       8,229ns │        3.1% │    3.01s │ 5,041 │
│        test_report_parsing[Rust ReadOnlyReport] │       8,815ns │        1.4% │    2.94s │ 4,828 │
│             test_report_parsing[EditableReport] │     195,443ns │        0.2% │    2.96s │ 1,005 │
│                 test_report_iterate_all[Report] │  47,338,292ns │        0.3% │    2.90s │    61 │
│         test_report_iterate_all[ReadOnlyReport] │  46,308,167ns │        1.7% │    2.95s │    63 │
│    test_report_iterate_all[Rust ReadOnlyReport] │  47,148,833ns │        1.3% │    2.94s │    62 │
│         test_report_iterate_all[EditableReport] │  45,321,666ns │        0.9% │    2.96s │    65 │
│              test_report_process_totals[Report] │  58,634,083ns │        0.7% │    2.97s │    50 │
│      test_report_process_totals[ReadOnlyReport] │  57,668,083ns │        0.4% │    2.90s │    50 │
│ test_report_process_totals[Rust ReadOnlyReport] │  58,458,584ns │        1.6% │    3.00s │    50 │
│      test_report_process_totals[EditableReport] │  55,609,125ns │        1.3% │    2.92s │    52 │
│                   test_report_filtering[Report] │ 178,210,958ns │        0.8% │    2.87s │    16 │
│           test_report_filtering[ReadOnlyReport] │ 184,516,500ns │        1.1% │    2.99s │    16 │
│      test_report_filtering[Rust ReadOnlyReport] │  74,746,791ns │        0.4% │    2.94s │    39 │
│           test_report_filtering[EditableReport] │ 170,324,791ns │        0.7% │    2.93s │    17 │
│                   test_report_serialize[Report] │       2,131ns │        2.6% │    2.97s │ 9,869 │
│           test_report_serialize[EditableReport] │   2,887,927ns │        1.5% │    2.98s │   256 │
│                       test_report_merge[Report] │ 239,321,583ns │        0.4% │    2.89s │    12 │
│               test_report_merge[EditableReport] │ 249,947,583ns │        0.2% │    2.76s │    11 │
│                test_report_carryforward[Report] │   2,933,698ns │        0.3% │    2.93s │   248 │
│        test_report_carryforward[EditableReport] │   2,941,503ns │        1.1% │    2.90s │   244 │
└─────────────────────────────────────────────────┴───────────────┴─────────────┴──────────┴───────┘

Comparing this with the numbers in #512, we can see that the process_totals benchmark got a speedup of 2x (-50ms), and the filtering benchmark as well got sped up by -50ms.

[codspeed-hq]

CodSpeed Performance Report

Merging #523 will create unknown performance changes

_{Comparing swatinem/consolidate-totals (0b4d0e3) with main (e6906af)}

Summary

⚠️ No benchmarks were detected in both the base of the PR and the PR.\

[giovanni-guidini]

LGTM, left some questions

[giovanni-guidini]

[nit] line_type and line.type (below) are confusing... Maybe we can rename line_type( something like line_coverage_type or line_coverage

If they are the same thing, why can't we match the block below in this match?

if line.type == "b":
    branches += 1
elif line.type == "m":
    methods += 1

[Swatinem]

I renamed the imports for the line_type and LineType above to make the distinction a little clearer.

[giovanni-guidini]

what is this?

[Swatinem]

A completely unused thing :-D Otherwise I don’t think anyone really knows.