Improve homepage search relevance

### Describe what's missing in our docs

The current homepage search does not consistently surface the most relevant or authoritative content for core concepts. For example, searching for “Inputs” (a fundamental Kestra concept) does not return the basic Inputs reference in the top results, but instead surfaces an older migration guide.

https://kestra.io/docs/workflow-components/inputs

We should improve the relevance and prioritization of search results for core concepts and reference documentation. This may involve improving our own indexing, ranking, or content weighting—or evaluating whether a third-party search solution (e.g. Algolia) would deliver better, more predictable results.

Goal: make it easy for users to quickly find the correct, up-to-date documentation for core Kestra concepts.

Start by reproducing the homepage search for “Inputs” and compare its results with the Inputs reference at https://kestra.io/docs/workflow-components/inputs. Investigate the existing indexing and ranking approach before evaluating whether a third-party search solution is needed; done means core concepts surface the correct, current reference documentation predictably.

Written by the indexing model from the issue text.