Update formatscaper according to identified requirements

During test runs, we saw that a "flat" assessment of endangerment status per file format ("endangered" vs. "safe") isn't enough. Even if a file format is at risk of becoming obsolete (or is already dead), this doesn't necessarily mean that a dataset containing such files is necessarily at risk (e.g. vim swap files likely just end up in the dataset accidentally and could be removed).

As such, we started following more of a risk management approach by evaluating the risk for each file, based on the probability of its format becoming obsolete, and the impact severity in case this file becomes uninterpretable.

formatscaper/models.py

 import dataclasses
 from typing import List, Optional
 
-from sqlalchemy import ForeignKey, create_engine
+from sqlalchemy import ForeignKey, UniqueConstraint, create_engine
 from sqlalchemy.orm import DeclarativeBase, Mapped, Session, mapped_column, relationship
 
 
@@ -98,6 +98,9 @@ class Result(ModelBase):
     format_id: Mapped[Optional[int]] = mapped_column(ForeignKey("format.id"))
     format: Mapped[Format] = relationship(back_populates="results")
 
+    # filenames are unique per record
+    __table_args__ = UniqueConstraint("record", "filename")
+
     @property
     def risk(self):
         """Calculate the risk assessment for the file."""