[Virtuoso461]

What does this PR do?

ExprExtractionConfig parsing dropped scientific-notation exponents: the number regex stopped right after the mantissa, so "1.5e3" was extracted as 1.5 and "6.02e23" as 6.02. Answers written in scientific notation were compared against the mantissa alone:

from math_verify import parse, verify
from math_verify.parser import ExprExtractionConfig

cfg = (ExprExtractionConfig(),)
parse("The answer is 1.5e3.", cfg)   # -> [1.5, '1.5']  (before this PR)
verify(parse("The answer is 1.5e3.", cfg), parse("The answer is 1500.", cfg))  # -> False

Fix

Adds an optional (?P<exponent>[eE][-+]?\d+) group after the integer/decimal alternation in number_re, and appends it when building the number string in extract_expr. Sympy's Number("1.5e3") then evaluates to 1500.

A trailing e without digits still does not count as an exponent — word1e extracts nothing, as before.

Tests

New tests/test_scientific_notation.py:

• positive / large / negative / uppercase exponents extract the full value
• 1.5e3 verifies equal to 1500
• no-false-positive control for trailing e

Full suite (excluding large corpus files): 241 passed, no regressions. On the unpatched tree the new tests fail 5/6.