Both looked for word in tags and content too, the only difference is
when searched for `#tag`, the prefix (`#`) was truncated before compared
with list of tags.
The previous commit broke this behaviour.
Looking for a tag means the union of **tags** AND **tag in content**, so
it has to search in the in currently found notes separetely, thus it has
to clone the list first (`.slice(0)`).
Some issues introduced in #954
(https://github.com/BoostIO/Boostnote/pull/954#issuecomment-336344915):
- search by `#tag1 #tag2` returns the results of only `#tag2`
- search by `#tag1 content` is as same as AND search by `#tag1 content` (and
`#tag1` is regarded a word, not a tag)
- search by `content #tag1` returns the results of only `#tag1`
This commit fixing these:
- search by `#tag1 #tag2` returns the results of only `#tag2`
- search by `#tag1 content` is as same as AND search by `#tag1 content` (and
`#tag1` is regarded a word, not a tag)
- search by `content #tag1` returns the results of only `#tag1`
NOTE: the examples works without `#` character too, because
`findByWord()` checks the tags too.
