WIP: Support constaints on distance column in KNN queries, for pagination and range queries

[asg017]

refs #165

.load dist/vec0
create virtual table vec_items using vec0(
  vector float[1]
);

insert into vec_items(rowid, vector)
  select value, json_array(value) from generate_series(1, 100);

select vec_to_json(vector), distance
from vec_items
where vector match '[1]'
  and k = 5;

/*
┌─────────────────────┬──────────┐
│ vec_to_json(vector) │ distance │
├─────────────────────┼──────────┤
│ '[1.000000]'        │ 0.0      │
│ '[2.000000]'        │ 1.0      │
│ '[3.000000]'        │ 2.0      │
│ '[4.000000]'        │ 3.0      │
│ '[5.000000]'        │ 4.0      │
└─────────────────────┴──────────┘
*/


select vec_to_json(vector), distance
from vec_items
where vector match '[1]'
  and k = 5
  -- the new magic
  and distance > 4.0;

/*
┌─────────────────────┬──────────┐
│ vec_to_json(vector) │ distance │
├─────────────────────┼──────────┤
│ '[6.000000]'        │ 5.0      │
│ '[7.000000]'        │ 6.0      │
│ '[8.000000]'        │ 7.0      │
│ '[9.000000]'        │ 8.0      │
│ '[10.000000]'       │ 9.0      │
└─────────────────────┴──────────┘
*/

TODO

• tests
• docs
• edge cases - when multiple items have same distance calculations
• should distances be f64?