While reviewing recent issues related to uncertainty handling and parsing behavior,
I noticed that invalid or partially invalid HGVS expressions appear to be handled
differently across variant types and code paths.

In some cases, parsing raises explicit exceptions, while in others it returns
partially parsed or normalized objects with implicit uncertainty. This makes it
difficult for downstream tools to reliably distinguish between:

• invalid input
• uncertain but valid representations
• valid but non-normalized expressions

Several existing issues touch on related aspects of this behavior (e.g. uncertainty
handling, implicit behaviors, and parsing edge cases), and I wanted to raise this
as a potential cross-cutting concern.

It may be useful to clarify the intended semantics, or to standardize how invalid
vs uncertain inputs are signaled across parsing and normalization steps.

I am happy to investigate a small, scoped improvement or clarification if this
aligns with the project’s direction.