perf: optimize scanl O(n²) --> O(n)

[walking-octopus]

append and last are both O(n), so calling them in a loop resulted in O(n²) time-complexity.

The PR replaces them with a plain loop that accumulates results using cons and reverses them.

I hope the code fits the expected style.

Benchmark

(time (void (scanl-ref + (range #e1e5))))
;; cpu time: 74668 real time: 75829 gc time: 11320

(time (void (scanl + (range #e1e5))))
;; cpu time: 2 real time: 2 gc time: 0

With a 100,000-element list, the new implementation is ~38,000x faster.

Tests

stas@thinkpad-t14 ~/P/racket-algorithms (scanl)> raco test main.rkt
raco test: (submod (file "main.rkt") test)
72 tests passed

[walking-octopus]

Inspired by Haskell, I also added a (scanl op seed xs) variant. I hope it's okay to do it in one PR. If you don't think that was a good idea, I can rebase and drop that commit. Thanks.

[codereport]

can you merge master into this PR so the ci.yml changes are here?

[codereport]

can you merge master into this PR so the ci.yml changes are here?

Yea, i just pushed a fix. It is working now #38