Prepare subset of noto (CJK and regular) based on another font

[jm4R]

I have a font (source_font), which contains regular characters and also glyphs from CJK characters. I wanted to change this font to noto. I've tried to modify some scripts to make it possible. What I've done was:

1. Subset regular noto based on source_font coverage -> subset_not_cjk.ttf
2. Converted CJK from OTF to TTF -> converted_cjk.ttf
3. Subset CJK based on source_font coverage -> subset_cjk.ttf
   So far it went OK. But when I've tried to merge subset_not_cjk.ttf with subset_cjk.ttf, different warnings and errors was shown:

WARNING:fontTools.merge:Dropped cmap subtable from font [1]:    format 14, platformID  0, platEncID  5
(...)
WARNING:fontTools.merge:Dropped mapping from codepoint 0XF904 to glyphId 'uniF904'
WARNING:fontTools.merge:Dropped mapping from codepoint 0XF908 to glyphId 'uniF908'
(... and many simillar warnings here ...)
Traceback (most recent call last):
  File "/some/path/noto_based_on.py", line 305, in <module>
    main()
  File "/some/path/noto_based_on.py", line 271, in main
    font = merger.merge(valid_files)
  File "/home/someuser/.local/lib/python3.6/site-packages/fontTools/merge.py", line 996, in merge
    table = clazz(tag).merge(self, tables)
  File "/home/someuser/.local/lib/python3.6/site-packages/fontTools/merge.py", line 159, in merge
    return m.mergeObjects(self, self.mergeMap, tables)
  File "/home/someuser/.local/lib/python3.6/site-packages/fontTools/merge.py", line 1044, in mergeObjects
    value = mergeLogic(getattr(table, key, NotImplemented) for table in tables)
  File "/home/someuser/.local/lib/python3.6/site-packages/fontTools/merge.py", line 48, in equal
    assert all(item == first for item in t), "Expected all items to be equal: %s" % lst
AssertionError: Expected all items to be equal: [NotImplemented, 'vhea']

The question is: Is this operation technically possible (assoming glyphs count doesn't exceed 2^16)? If so, why the process shows such warnings/errors?