feat(recommender): ✨ initial sketch of the activity recommender

[dvorka]

Tasks:

• as expected, vector-based activity (w/ possible feature embeddings add-on) lookup does not work quite well, it's rather useful to have some fun. Not sure whether I will have any use for it, except catchy cluster chart.
• instead of vector/cosine design a reasonable valuation function / metric which will bring the most similar activities based on the time/distance/elevation delta (properly weighted).

---

This PR brings initial sketch of the activity recommender.

[Copilot]

Pull request overview

Adds an initial activity recommender feature to MyTraL, exposing a new UI page that ranks a user’s historical activities by similarity to a target duration/distance/elevation, and optionally visualizes clustered matches.

Changes:

• Introduces a pure-numpy recommender module (cosine similarity over normalized features) plus clustering/PCA support.
• Adds a new /app/recommender route, form, and UI template, and wires it into the navigation and app startup.
• Adds Bokeh scatter plotting for clustered matches and a dedicated pytest suite covering the recommender logic.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
mytral/recommender.py	Core recommendation + clustering logic (normalize, cosine similarity, k-means, PCA projection).
mytral/blueprints/recommender_uri_space.py	New route handling form submission, running recommender, and preparing chart data.
mytral/forms.py	Adds ActivityRecommenderForm (filters + numeric query inputs).
mytral/templates/activity-recommender.html	New page UI for inputs, results table, and scatter plot container.
mytral/charts.py	Adds Bokeh scatter plot renderer for clustered recommender matches.
mytral/views.py	Adds ActivityRecommendationView adapter for template-friendly rendering of matches.
mytral/templates/layout.html	Adds a navigation entry to reach the recommender page.
mytral/run.py	Registers the recommender route module in the web entrypoint.
mytral/run_desktop.py	Registers the recommender route module in the desktop entrypoint.
tests/test_recommender.py	New unit tests for filtering, normalization, ranking, type voting, and clustering behavior.