An empty `page_range` is treated as 'all pages' instead of 'no pages'

[YizukiAme]

Bug Description

load_pdf() branches on if page_range: rather than page_range is not None. An explicitly empty selection such as [] or range(0) falls into the else branch and is replaced with list(range(last_page)) — processing the entire PDF instead of zero pages.

Programmatic callers using load_pdf(), load_from_file(), load_from_folder(), or CLILoader with a precomputed empty page selection get the opposite of what they asked for.

Root Cause

surya/input/load.py L19-27:

def load_pdf(pdf_path, page_range: List[int] | None = None, dpi=settings.IMAGE_DPI):
    doc = open_pdf(pdf_path)
    last_page = len(doc)

    if page_range:      # ← [] is falsy, falls through!
        assert all([0 <= page < last_page for page in page_range])
    else:
        page_range = list(range(last_page))  # ← processes ALL pages

Steps to Reproduce

from surya.input.load import load_pdf

# User explicitly requests zero pages
images, names = load_pdf("document.pdf", page_range=[])
print(len(images))  # Expected: 0, Actual: len(document) — all pages!

Expected Behavior

page_range=[] should result in no pages being processed, or should raise a clear ValueError.

Suggested Fix

if page_range is not None:  # distinguish "not provided" from "empty"
    if len(page_range) == 0:
        return [], []  # or raise ValueError("Empty page range")
    assert all([0 <= page < last_page for page in page_range])
else:
    page_range = list(range(last_page))