feat: add batcave.biz support, closes #6

## What changed

- `BatcaveBizMarkup` now accepts a `clientChan chan *http.Client` and
  sends the authenticated cookie jar client back to the caller after
  completing the Cloudflare challenge flow. All error paths send nil so
  the caller never blocks.

- `Comic` struct gains a `Client *http.Client` field. `NewComic` wires
  up the channel, receives the client, and stores it so downstream code
  can reuse the same authenticated session.

- `downloadFile` branches on `c.Client`: when set it builds the request
  manually and only attaches a `Referer: https://batcave.biz/` header
  when the image URL is actually on batcave.biz. Some issues host images
  on third-party CDNs (e.g. readcomicsonline.ru) that actively block
  requests with a batcave Referer, returning 403 — omitting the header
  fixes those.

- `ParseBatcaveBizTitle` extracts the chapter title from the
  `__DATA__.chapters` JSON array by matching the chapter ID in the URL's
  last path segment. The HTML `<title>` on batcave.biz is prefixed with
  "Read " and suffixed with "comics online for free", making it
  unsuitable as a filename. Using the chapter data gives clean titles
  like "Nightwing (1996) 153". "Issue #" and bare "#" are stripped since
  the hash character causes problems on some filesystems and tools.

- `ParseBatcaveBizImageLinks` now unescapes `\/` → `/` in extracted
  URLs. The `__DATA__` JSON often contains forward-slash-escaped URLs
  that would otherwise be stored verbatim.

- `archive.go`: `filepath.Walk` was called on `filepath.Dir(sourcePath)`
  (the library root) instead of `sourcePath` (the comic's own folder).
  This caused any leftover image files from previous downloads in sibling
  directories to be included in every new CBZ. Fixed by walking
  `sourcePath` directly.

- `BatcaveBizMarkup` client now has a 30s `Timeout`. Without it, a
  single stalled CDN connection would hang the worker goroutine
  indefinitely, causing `Download()` to block forever waiting for a
  result that never arrives.

- Fixed `for e := range err` in `cli/root.go` — ranging over `[]error`
  with one variable yields the index, not the error value.

comic/download.go

@@ -6,6 +6,7 @@ import (
 	"net/http"
 	"os"
 	"path/filepath"
+	"strings"
 	"time"
 
 	cloudflarebp "github.com/DaRealFreak/cloudflare-bp-go"
@@ -39,13 +40,33 @@ func downloadFile(url string, page int, c *Comic) error {
 		}
 	}
 
-	res, err := handleRequest(url)
+	var res *http.Response
+	var err error
+	if c.Client != nil {
+		req, reqErr := http.NewRequest("GET", url, nil)
+		if reqErr != nil {
+			return ComicDownloadError{Message: "invalid request", Code: 1}
+		}
+		req.Header.Set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36")
+		if strings.Contains(url, "batcave.biz") {
+			req.Header.Set("Referer", "https://batcave.biz/")
+		}
+		res, err = c.Client.Do(req)
+	} else {
+		res, err = handleRequest(url)
+	}
 	if err != nil {
 		return ComicDownloadError{
 			Message: "invalid request",
 			Code:    1,
 		}
 	}
+	if res.StatusCode != http.StatusOK {
+		return ComicDownloadError{
+			Message: "bad response",
+			Code:    1,
+		}
+	}
 	defer res.Body.Close()
 
 	imageFile, err := os.Create(imageFilepath)