Fix unreachable error check in Mnemonic.to_entropy()

[mapaba79]

Problem

to_entropy() looks up each word's index with list.index():

ndx = self.wordlist.index(self.normalize_string(word))
if ndx < 0:
    raise LookupError('Unable to find "%s" in word list.' % word)

list.index() never returns a negative value — when the element is
not found, it raises ValueError directly instead of returning -1.
As a result, the if ndx < 0 branch is unreachable: it can never be
true, because if the word isn't in the wordlist the exception is
already raised two lines earlier, before the check even runs.

Impact

When an invalid/misspelled word is passed to to_entropy(), callers
get a raw ValueError with Python's generic list-lookup message
(e.g. 'notaword' is not in list) instead of the intended,
descriptive LookupError (Unable to find "notaword" in word list.). This is misleading for end users debugging a bad mnemonic,
and breaks any caller that specifically catches LookupError
(matching how the rest of this library reports lookup failures)
without also catching ValueError.

Reproduction

from mnemonic import Mnemonic
m = Mnemonic("english")
m.to_entropy("abandon " * 11 + "notaword")
# ValueError: 'notaword' is not in list

Fix

Wrap the .index() call in a try/except ValueError and raise the
intended LookupError from there, instead of checking a condition
that can never be true:

try:
    ndx = self.wordlist.index(self.normalize_string(word))
except ValueError:
    raise LookupError('Unable to find "%s" in word list.' % word)

Testing

• Reproduced the original issue against a clean checkout: confirmed
  a raw ValueError was raised instead of LookupError.
• Applied the fix and re-ran the same reproduction: confirmed
  LookupError: Unable to find "notaword" in word list. is now
  raised as intended.
• Ran the full existing test suite (pytest tests/) before and
  after the change: all 7 tests pass in both cases, confirming no
  regression to existing checksum/entropy behavior.

This is a pure error-handling fix; the entropy/checksum computation
logic is unchanged.