The tricky part comparing to those existing flags is that their
behavior is defined by a single character, which is how
_isAutoClosingOpenCharType() works. The comments, on another hand,
can consists of multiple characters and they do not fit existing
auto-closing filtering.

To avoid duplication of tokenizing, the decision on whether to place
closing comment or not is based on the electric action which now stored
corresponding token which generated new text.

This allows to filter electric actions in cursor type operations.

Similar to opening/closing pairs the opening/closing comment blocks
are stores in a character map.

Could be a bit overkill, but could as well be more flexible in case
language configuration supports multiple opening/closing comment
blocks (not sure if it is already supported, but in Pascal it is
possible to use { and (* for block comments).

P.S. There might be a trick to fit this logic into existing
isAutoClosing functionality. It does perform tokenization at certain
conditions, but the early output checks and checks for auto-closing
pairs would need to be changed somehow. And is still needed to know
whether typing sequencer is a quote, comment or a brace.
Probably, can check if it's a quote or a brace and consider a
comment otherwise, but seems a bit weak.