[CLI] Feature — Local Search and Recall

[0xheartcode]

Overview

Developers often vaguely remember an issue — "that authentication bug from last week" — but finding it again means opening a browser and searching through potentially dozens of projects. If you work across multiple providers, there's no single search that spans both.

Related to but different scope from #4 (which covers TUI fuzzy filtering within a board view). This issue covers CLI-level search against both live API and local cache.

Current State

No search or recall functionality. No browsing history or bookmarks.

Proposed Solution

Add search and recall capabilities against both the live API and local cache:

lazydash search "authentication bug"          # live API search
lazydash search --label "priority:high"       # filtered live search
lazydash search --assignee "@me" --state open # filtered live search
lazydash recall "that auth thing"             # fuzzy search LOCAL cache only
lazydash recent                               # last N items you viewed
lazydash bookmarks                            # items you explicitly starred

• search hits the provider API in real-time across all configured providers
• recall searches only local cache — works offline, instant, fuzzy matching
• recent shows browsing history (tracked automatically as you navigate the TUI)
• bookmarks shows items explicitly marked as important (via b keybinding in TUI)

Because the cache stores normalized types, all search/recall works across all providers.

Priority

Do second, alongside the cache — they're deeply intertwined.

Acceptance Criteria

• lazydash search queries live API across all configured providers
• lazydash recall performs fuzzy search against local cache only
• lazydash recent shows browsing history
• lazydash bookmarks shows explicitly starred items
• Search supports filters (--label, --assignee, --state)
• Search works across all providers in unified results

Epic: This is an epic-sized issue — covers multiple CLI commands and integrates tightly with the cache layer. Will need to be broken down into smaller sub-issues when work begins.