fix: Handle empty strings in StrictWhitespaceControlParser

[rewaddell]

What

Add empty-string guards to StrictWhitespaceControlParser so hasLeftTrim,
hasRightTrim, stripLeft, and stripRight no longer index into an empty
token.

Why

hasLeftTrim called unwrapped.charAt(0) and hasRightTrim called
unwrapped.charAt(unwrapped.length() - 1) with no empty-string guard. Any
template parsed with withParseWhitespaceControlStrictly(true) that produces
an empty unwrapped token (e.g. {{}}) threw StringIndexOutOfBoundsException
during parsing.

The lenient parser already handles this gracefully — WhitespaceUtils.startsWith
/ endsWith return false for null/empty input. This change gives the strict
parser the same defensive guarantee, so the two implementations behave
consistently on empty tokens.

Fixing it here protects every consumer of the strict parser rather than
sanitizing input in a single downstream call path.

Behavior change

• {{}} now renders to an empty string instead of throwing.
• An empty tag token ({%%}) still errors, but with the expected Unknown tag
  error rather than a StringIndexOutOfBoundsException — correct downstream
  behavior, unchanged in intent.

Tests

• StrictWhitespaceControlParserTest: empty token returns false from the
  hasX methods and is returned unchanged by the stripX methods; non-empty
  trim markers still detected and stripped.
• LegacyWhitespaceControlParsingTest.itHandlesEmptyExpressionToken: {{}}
  renders to "" in strict mode.

---

PR description authored by Claude Code.