Copying characters on supplementary planes is broken

[a4lg]

Attach (recommended) or Link to PDF file here: sample PDF file

Configuration:

• Web browser and its version: Firefox 65.0 (64-bit)
• Operating system and its version: Windows 10 1810 (64-bit)
• PDF.js version: 2.0.943 (pdf.js prebuilt) and 2.1.97 (browser extension in Firefox 65.0)
• Is a browser extension: Yes and No (see above)

Steps to reproduce the problem:

1. Open sample PDF file
2. Select all (Ctrl+A) and copy all the text (Ctrl+C)
3. Paste to some text editor (with proper Unicode and emoji support)

What is the expected behavior? (add screenshot)

Following text (without notes) is copied:

OK: ☺ (Note: U+263A)
NG: 😃 (Note: U+1F603)
In the following example, hidden text (not supposed to be copied) is encoded:
𠁔𠁨𠁩𠁳𠀠𠁩𠁳𠀊𠁮𠁯𠁴𠀉𠁷𠁨𠁡𠁴𠀠𠁉𠀠𠁷𠁡𠁮𠁴𠁥𠁤𠀮
(Note: garbled Chinese text which I encoded a hidden string as U+00xx -> U+200xx)

What went wrong? (add screenshot)

Following text (without notes) is copied (Note: formatted for readability):

OK: ☺ (Note: U+263A)
NG:  (Note: U+0F603)
In the following example, hidden text (not supposed to be copied) is encoded:
This is not what I wanted.

Link to a viewer (if hosted on a site other than mozilla.github.io/pdf.js or as Firefox/Chrome extension):

I have a fix and will pull-request later.