perf(MergedText): if all chunks are merged, just concatenate them all #285

MergedText/early-return/compact
                        time:   [384.24 ns 385.19 ns 386.23 ns]
                        change: [-98.397% -98.364% -98.332%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  4 (4.00%) high mild
  2 (2.00%) high severe
MergedText/early-return/line_based
                        time:   [394.09 ns 395.57 ns 397.52 ns]
                        change: [-23.201% -22.035% -20.748%] (p = 0.00 < 0.05)
                        Performance has improved.

I'd say this scenario is realistic in practice, since sometimes we do manage to solve all the conflicts. This is my first `perf` commit to use criterion, and the results are quite impressive :) ``` MergedText/early-return/compact time: [384.24 ns 385.19 ns 386.23 ns] change: [-98.397% -98.364% -98.332%] (p = 0.00 < 0.05) Performance has improved. Found 6 outliers among 100 measurements (6.00%) 4 (4.00%) high mild 2 (2.00%) high severe MergedText/early-return/line_based time: [394.09 ns 395.57 ns 397.52 ns] change: [-23.201% -22.035% -20.748%] (p = 0.00 < 0.05) Performance has improved. ``` I would be glad to attach the benchmark to the PR, but it was done in a somewhat hacky way -- since the benchmarks are compiled as external packages, I needed to make a bunch of stuff public so that they could access it