Add explicit CDATA guard cleanup

[Joseph-Mutua]

Summary

• add Node#unwrap_cdata_text as an explicit, non-mutating convenience API
• remove bare, line-comment, block-comment, and nested block-comment CDATA guards only when they form one complete outer pair
• preserve text / content, surrounding whitespace, and incomplete, mismatched, or embedded guards unchanged
• compose with Ruby's String#strip as node.unwrap_cdata_text.strip when generic surrounding-whitespace cleanup is wanted
• use readable, CDATA-specific guard patterns with greedy outer-body matching
• cover every guard form discussed in the issue plus boundary, composition, exact-whitespace, and immutability cases

API choice

The CDATA-specific name reflects the method's deliberately narrow responsibility. Whitespace cleanup remains the responsibility of String#strip, so it behaves consistently for guarded and ordinary text.

Validation

• bundle exec ruby -Itest -Ispec spec/node_spec.rb — 138 runs, 328 assertions
• bundle exec rake test — 312 runs, 4,660 assertions
• generated guard matrix — 10,000 cases, 0 failures
• bundle exec yardoc --no-output lib/nokolexbor/node.rb

@zyc9012 @forthrin, this follows the explicit opt-in direction from #31 while keeping normal DOM accessors lossless.

Fixes #31

[zyc9012]

Thanks @Joseph-Mutua!

[zyc9012]

I think the method name is too general.

Maybe unwrapped_cdata_text or unwrap_cdata_text

[Joseph-Mutua]

Addressed in aac2c60: the method is now Node#unwrap_cdata_text, matching its CDATA-specific scope.

I also separated the two transformations noted on the issue. The helper now removes only a complete recognized CDATA guard and preserves whitespace; callers can use node.unwrap_cdata_text.strip when they also want generic surrounding-whitespace cleanup.

[zyc9012]

Thank you @Joseph-Mutua! LGTM.