Feat: adds facets to search

[micheleriva]

This PR aims to add support for facets while searching with Lyra.
The API is pretty straightforward and will likely close #206, and pave the ground for #212.

Example API

Given the following database, we'll now need to select which fields will be available while performing a facet query:

const db = await create({
  schema: {
    id: "string",
    brand: "string",
    model: "string",
    year: "number",
    price: "number",
    used: "boolean",
    available: "boolean",
  }
});

Now let's insert some data. I'm experimenting with a format similar to the following one:

{"id":"3e4e0dc3-7cb9-4a03-9a36-2c66f51e08c9","brand":"Acura","year":2000,"model":"TL","available":true,"used":false,"price":36899},
{"id":"321b5c59-be55-48b9-a3d4-3e1586d1a863","brand":"Dodge","year":2009,"model":"Journey","available":false,"used":true,"price":12799},
{"id":"62123f1e-9f13-474a-a282-8d141c296edf","brand":"Nissan","year":1998,"model":"Frontier","available":true,"used":false,"price":48840},

So let's batch insert using the batchInsert built-in function:

await insertBatch(db, data);

Now we need to perform search. Lyra currently supports 3 data types: string, boolean, number, and we will be able to perform facet queries on all of them:

const results = await search(db, {
  term: "Dodge challenger",
  facets: {
    brand: {
      size: 10,
      sort: "ASC",
    },
    model: {
      size: 10,
      sort: "ASC",
    },
    available: {
      size: 10,
    },
    year: {
      ranges: [
        { from: 1990, to: 1995 },
        { from: 2000, to: 2005 },
        { from: 2005, to: 2010 },
        { from: 2010, to: 2015 },
        { from: 2015, to: 2020 },
      ]
    }
  },
  limit: 2,
})

This will lead to the following result:

{
  elapsed: 950750n,
  hits: [
    {
      id: 'a9c2a700-c23d-42d1-921d-99442d0d6af5',
      score: 8.586111395567093,
      document: {
        id: 'a9c2a700-c23d-42d1-921d-99442d0d6af5',
        brand: 'Mitsubishi',
        year: 1999,
        model: 'Challenger',
        available: false,
        used: true,
        price: 31861
      }
    },
    {
      id: '321b5c59-be55-48b9-a3d4-3e1586d1a863',
      score: 3.6680631599644267,
      document: {
        id: '321b5c59-be55-48b9-a3d4-3e1586d1a863',
        brand: 'Dodge',
        year: 2009,
        model: 'Journey',
        available: false,
        used: true,
        price: 12799
      }
    },
  ],
  count: 52,
  facets: {
    brand: { count: 2, values: { Dodge: 51, Mitsubishi: 1 } },
    model: {
      count: 26,
      values: {
        Viper: 4,
        Stratus: 4,
        'Ram 2500': 4,
        'Grand Caravan': 4,
        Dakota: 4,
        'Ram 3500': 4,
        'Dakota Club': 3,
        Journey: 2,
        Charger: 2,
        Nitro: 2
      }
    },
    available: { count: 2, values: { true: 27, false: 25 } },
    year: {
      count: 4,
      values: {
        '2000-2005': 15,
        '2005-2010': 13,
        '1990-1995': 10,
        '2010-2015': 5
      }
    }
  }
}

This is a work in progress

This PR is a huge work in progress. Types are not working at their best yet, and algorithms are not optimized at all. We should also return a list of IDs to perform a "group by" operation once a user decides to enter a facet.

Please do not review yet.