`gte` is 1000+ times slower than plain filter

[mtsmfm]

Describe the bug

search(db, { where: { n: { gte: 1 } } }) takes 500ms but numbers.filter((n) => n >= 1).length can be finished within 0.3ms

To Reproduce

Run the following snippet.

import { create, insertMultiple, search } from "@orama/orama";

const N = 10000;
const main = async () => {
  const numbers = Array.from({ length: N }, (_, i) => i);

  const db = create({
    schema: {
      n: "number",
    },
  });
  insertMultiple(
    db,
    numbers.map((n) => ({ n })),
    numbers.length
  );

  console.time("search");
  const result1 = await search(db, { where: { n: { gte: 1 } } });
  console.timeEnd("search");

  console.time("filter");
  const result2 = numbers.filter((n) => n >= 1).length;
  console.timeEnd("filter");

  if (result1.count !== N - 1) {
    throw new Error("invalid count");
  }
  if (result2 !== N - 1) {
    throw new Error("invalid count");
  }
};

main();

/*
$ NODE_NO_WARNINGS=1 node --experimental-strip-types test.ts
search: 588.004ms
filter: 0.241ms
*/

Expected behavior

search should be finished at the latest the same-ish speed. Considering Orama can create sorted index, it's preferable to be finished faster than entire search though.

Environment Info

System:
    OS: Linux 5.15 Ubuntu 22.04.4 LTS 22.04.4 LTS (Jammy Jellyfish)
    CPU: (12) x64 AMD Ryzen 5 3600 6-Core Processor
    Memory: 8.59 GB / 15.57 GB
    Container: Yes
  Binaries:
    Node: 22.13.0 - ~/.local/share/mise/installs/node/22/bin/node
    npm: 11.1.0 - ~/.local/share/mise/installs/node/22/bin/npm

Affected areas

Search

Additional context

No response

[micheleriva]

fixed in #973, thanks for spotting this. It may still be a bit slower on little datasets, but the speed improvement will be noticeable when working on larger datasets since the native filter method has O(n) time complexity vs O(log n) of AVL tree search