fix: unescape HTML entities in result titles

Wikipedia returns HTML entities like <span> which were being
double-escaped by Go templates. Now using html.UnescapeString and
template.HTML to render properly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

internal/views/views.go

@@ -19,6 +19,7 @@ package views
 import (
 	"embed"
 	"encoding/xml"
+	"html"
 	"html/template"
 	"io/fs"
 	"net/http"
@@ -70,6 +71,10 @@ type ResultView struct {
 	// TemplateName is the actual template to dispatch to, computed from Template.
 	// "videos" maps to "video_item", everything else maps to "result_item".
 	TemplateName string
+	// SafeTitle and SafeContent are HTML-unescaped versions for rendering.
+	// The API returns HTML entities which Go templates escape by default.
+	SafeTitle   template.HTML
+	SafeContent template.HTML
 }
 
 // PageNumber represents a numbered pagination button.
@@ -205,7 +210,12 @@ func FromResponse(resp contracts.SearchResponse, query string, pageno int, activ
 			r.URL = &safe
 		}
 		r.Thumbnail = util.SanitizeResultURL(r.Thumbnail)
-		pd.Results[i] = ResultView{MainResult: r, TemplateName: tmplName}
+		pd.Results[i] = ResultView{
+			MainResult:   r,
+			TemplateName: tmplName,
+			SafeTitle:    template.HTML(html.UnescapeString(r.Title)),
+			SafeContent:  template.HTML(html.UnescapeString(r.Content)),
+		}
 	}
 
 	// Convert answers (they're map[string]any — extract string values).