SkPDF:  Implement /ActualText to make text extraction correct.

For old API: no change.

For new API: LTR text is perfectly extracted, RTL needs better testing.

CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-Clang-GCE-CPU-AVX2-x86_64-Debug-ASAN-Trybot,Test-Ubuntu-Clang-GCE-CPU-AVX2-x86_64-Debug-MSAN-Trybot

BUG=skia:5434
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2322403002

Review-Url: https://codereview.chromium.org/2322403002
3 files changed