Mark EmptySmallCircle, EmptySet and eight other shapes as letter-like

[tombouctou]

WMA documents these named characters as "Letter-like form":

character	WMA reference
\[EmptyCircle]	https://reference.wolfram.com/language/ref/character/EmptyCircle.html
\[EmptyDiamond]	https://reference.wolfram.com/language/ref/character/EmptyDiamond.html
\[EmptyDownTriangle]	https://reference.wolfram.com/language/ref/character/EmptyDownTriangle.html
\[EmptyRectangle]	https://reference.wolfram.com/language/ref/character/EmptyRectangle.html
\[EmptySet]	https://reference.wolfram.com/language/ref/character/EmptySet.html
\[EmptySmallCircle]	https://reference.wolfram.com/language/ref/character/EmptySmallCircle.html ("Not the same as the infix operator \[SmallCircle]")
\[EmptySmallSquare]	https://reference.wolfram.com/language/ref/character/EmptySmallSquare.html
\[EmptySquare]	https://reference.wolfram.com/language/ref/character/EmptySquare.html
\[EmptyUpTriangle]	https://reference.wolfram.com/language/ref/character/EmptyUpTriangle.html
\[FilledDiamond]	https://reference.wolfram.com/language/ref/character/FilledDiamond.html

In named-characters.yml they had is-letter-like: false and are not operators either, so the tokeniser rejected them wherever they appeared. For example, in Mathics3:

In[1]:= Hold[Infix[t, \[EmptySmallCircle]]]
Syntax::sntxf: "Hold[Infix[t, \[EmptySmallCircle]" cannot be followed by "]]".

This PR sets is-letter-like: true for these ten characters.

A wider note: about 215 named characters are currently neither letter-like nor operators, so they can't be used in an expression. Many of them are legitimately so (spaces, Raw* characters, brackets), but some others may also be letter-like in WMA (e.g. Euro, Cent, the chess pieces). I only changed the ten above, which I checked one by one against the WMA reference pages.

Tests: test_letterlike_geometric_shapes in test/test_tokeniser.py.

🤖 Generated with Claude Code

[rocky]

LGTM - thanks for finding this.