Validate uppercase exponent underscore placement

[0xjc65eth]

Summary

• reject decimal literals with underscores adjacent to uppercase E exponent markers
• extend the existing decimal underscore syntax test with 12_E34 and 12E_34

Why
The scanner accepts both e and E exponent markers, but the post-lex underscore checks only searched for _e and e_. This let malformed uppercase forms bypass the same validation applied to lowercase forms.

Testing

• python3 offset check for the updated syntax test expectations
• scripts/error_codes.py --check
• scripts/check_style.sh libsolidity/analysis/SyntaxChecker.cpp test/libsolidity/syntaxTests/parsing/lexer_numbers_with_underscores_decimal_fail.sol
• git diff --check

Closes #16740.