What kind of change does this PR introduce?

Bug fix

What is the current behavior?

Named HTML entities containing digits (e.g. ¹, ½, ▓, ∴) are not processed correctly by the Angular template compiler. Instead of being decoded to their corresponding Unicode characters, they are output as escaped text (e.g. ¹ instead of ¹).

This affects all 24 valid HTML named entities that contain digits in their names: blk12, blk14, blk34, emsp13, emsp14, frac12, frac13, frac14, frac15, frac16, frac18, frac23, frac25, frac34, frac35, frac38, frac45, frac56, frac58, frac78, sup1, sup2, sup3, there4.

Closes #51323

What is the new behavior?

The lexer's isNamedEntityEnd function now accepts digits as valid characters within named entity references, allowing all standard HTML named entities to be parsed and decoded correctly.

Additional context

The root cause was in the isNamedEntityEnd function in packages/compiler/src/ml_parser/lexer.ts. This function determines when to stop scanning a named entity reference. It used !chars.isAsciiLetter(code) as its termination condition, which meant any digit would immediately end the entity name scan. For an entity like ¹, the scanner would stop at sup (before the 1), never reach the semicolon, and fall back to treating the & as plain text.

The fix adds chars.isDigit(code) to the valid character check, so entity names like sup1 and frac12 are fully scanned.

The entity lookup table in entities.ts already contained all 24 entities — only the lexer scanning logic needed to be updated.