0.40.0

• Add Python 3.15 support.
• Drop Python 3.10 support.
• Fix arbitrary code execution: exclude pickle from format auto-detection, it must be requested explicitly with format="pickle" or from_pickle. (GHSA-v3xm-c46x-phwc) (by @Sengtocxoen)
• Fix match string patterns: match special characters literally (only * is a wildcard) and prevent catastrophic backtracking (ReDoS). #624
• Fix match string patterns matching keys with a trailing newline. (by @fhgffy in #619)
• Include tuple indexes in keypaths discovery and match. (by @xiehuanyi in #620)
• Allow frozen dicts to use methods returning a new dict (flatten, unflatten, subset, filter, invert). (by @itzzdev09 in #614)
• Fix flatten quadratic time complexity (up to ~25x faster on nested data).
• Fix flatten with indexes=True on benedict instances: flattened keys (eg. a[0]) were parsed as keypaths.
• Fix BaseDict.setdefault raising AssertionError with omitted or None default. (by @harbinresearcher in #626)
• Fix keyword arguments ignored when constructing from a mapping (eg. benedict(d, key=value)), they are now set on the wrapped dict. (by @sricharanreddycheruku in #627)
• Fix remove with a single non-string key (eg. remove(0)). (by @xiehuanyi in #628)
• Fix BaseDict.dict and union operator type annotations to return a plain dict. (by @xiehuanyi in #629)
• Fix rename and remove with deep=True to traverse nested lists and tuples. (by @xiehuanyi in #616)
• Fix base64 serializer to keep the Base64 representation ASCII with non-UTF-8 payload encodings. (by @xiehuanyi in #615)
• Fix ini decoding of valueless options when using allow_no_value=True. (by @ac1982 in #618)
• Fix ini encoding of valueless [DEFAULT] options when using allow_no_value=True.
• Fix csv decoding of strings using carriage-return line endings. (by @fhgffy in #621)
• Fix csv encoding losing values of non-string keys. (by @fhgffy in #622)
• Fix xls decoding of sheets without columns row and with numeric column names. (by @fhgffy in #623)
• Bump requirements.