Optimizing client-side search weighting

[jasalt]

Follow up on #34

Testing the search functionality typing search query "interop", the weighting might have some room for improvement, while there's no critical issue and the index seems quite complete.

in

• Host Interop > Instance checks compose
• Infix Notation > Parsing strings at runtime: from-string > The environment

Observation: I click "Host Interop" expecting to navigate to the beginning of the page, but end up into sub-heading.

int

• Native Interop (FFI) > Binding a function
• Native Interop (FFI) > Binding a function > :blocking
• Foreign Functions (jolt.ffi) > Binding functions
• Native Interop (FFI) > Binding a function > :& (variadic C functions)
  

Observation: Same, I click "Native Interop (FFI)" expecting to navigate to the beginning of the page, but end up into sub-heading.

inter

• RFC 0005: Structural Type Interference > Inference rules
• RFC 0005: Structural Type Interference > Soundness
• RFC 0004: Type Hints > Status and non-goals
• RFC 0005: Structural Type Interference > The type lattice > Termination: deth cap

Observation: Where did the previous results go?

intero

• no results

Observation: Where did all the results go?

interop

• Host Interop
• Cljc Interop
• Native Interop (FFI)
• Host Interop

Observation: These are what I expected.

Conclusion

Typing "interop" the search gives slightly surprising results with a partial search query, often pointing to quite deep in sub-headings, while the final result is fine.

With partial results the sub-headings seem to be getting prioritized too much over top-level headings / page titles.

Worth checking also why "intero" doesn't return any results.

I didn't look at the implementation but these are some observations / guesses.