merge version 0.7.0 from upstream

[brandonstubbs]

This is the initial part in order to solve BetterThanTomorrow/calva#807

I have merged an "as-is" approach the indentation changes on top of the latest 0.7.0 version.
While doing this I can see there is a lot of history discussed around this topic: weavejester#77

The tests have purposely been left as failing to keep them open as discussion points. Some will fail as the indentation is applied.
However, see others below:

Do we want to remove the comma in this case?

# case 1
FAIL in (test-remove-multiple-non-indenting-spaces) (core_test.cljc:638)
expected: ["{a, b}"]
  actual: ["{a b}"]

We should probably remove the comma in this case, or keep it like the next case?

# case 2
FAIL in (test-options) (core_test.cljc:1001)
expected: ["{:one two, " " :three four}"]
  actual: ["{:one   two" " , " " :three four}"]

# case 3
FAIL in (test-options) (core_test.cljc:1007)
expected: ["{:one two," " :three four}"]
  actual: ["{:one   two," " :three four}"]

Another comma case:

# case 4
FAIL in (test-surrounding-whitespace) (core_test.cljc:676)
surrounding whitespace removed
expected: ["{:x 1, :y 2}"]
  actual: ["{:x 1" " , :y 2}"]

Then I would like to add the case from BetterThanTomorrow/calva#807 which is the alignment when there is type hinting:

# case 5
(is (reformats-to?
       ["(let [^String x \"foo\""
        "      some-longer-var \"bar\"])"]
       ["(let [^String x       \"foo\""
        "      some-longer-var \"bar\"])"]
       {:split-keypairs-over-multiple-lines? true}))

FAIL in (test-options) (core_test.cljc:886)
expected: ["(let [^String x       \"foo\"" "      some-longer-var \"bar\"])"]
  actual: ["(let [^String x               \"foo\"" "      some-longer-var \"bar\"])"]

[PEZ]

Hello. Meny thanks for working with this!

It is a bit late here and I am too tired to make much sense of anything, so will have to get back to this tomorrow. 😄

Generally for comma cases, I think that default behaviour should be to never add or remove commas. (Not sure what the failing tests means regarding, since I don't see the input.)

Also generally. I think Calva's cljfmt should behave as much as possible as the current 0.7.0 one. I can from the top of my head not come up with any other exception than this alignment option, which is the only reason we use a fork.

I realise those two general observations could be in conflict. Are they?