Add advisory locking to mongodb

[andyN42]

Resolves #435

This implements advisory locking for mongodb. The migrate_advisory_lock collection will have a unique index on the locking_key collection. Migrate processes will attempt to claim the lock by inserting the same unique value into the collection. If succesful, they have the lock, if not, they wait for up to 15s using an exponential backoff.

Note that the collection name and backoff max time are configurable with the following query params:

URL Query	WithInstance Config	Description
x-advisory-locking	true	Feature flag for advisory locking, if set to false, disable advisory locking
x-advisory-lock-collection	migrate_advisory_lock	The name of the collection to use for advisory locking
x-advisory-lock-timout	15	The max time in seconds that the advisory lock will wait if the db is already locked.
x-advisory-lock-timout-interval	10	The max timeout in seconds interval that the advisory lock will wait if the db is already locked.

[coveralls]

Pull Request Test Coverage Report for Build 870

• 95 of 134 (70.9%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.4%) to 54.246%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
database/mongodb/mongodb.go	95	134	70.9%

Totals
Change from base Build 863:	0.4%
Covered Lines:	2868
Relevant Lines:	5287

---

💛 - Coveralls

[dhui]

Thanks for the PR!

[andyN42]

@dhui I am seeing intermittent cassandra test failures, unrelated to my changes. Is this a known issue?

https://travis-ci.com/github/golang-migrate/migrate/jobs/391121339

    --- PASS: TestMigrate/cassandra:3.0 (26.99s)
    --- PASS: TestMigrate/cassandra:3.11 (27.69s)
FAIL
coverage: 49.3% of statements
exit status 1
FAIL	github.com/golang-migrate/migrate/v4/database/cassandra	78.845s
?   	github.com/golang-migrate/migrate/v4/database/clickhouse	[no test files]
=== RUN   Test

Edit: force pushed whitespace changes to my branch twice. First time, same issue, second time, build passed

[dhui]

Thanks for addressing the prev round of feedback.

Re: failing tests
Yeah, unfortunately, some of the integration tests fail intermittently... e.g. docker pull fails or container fails to start 😞
Force pushing is the best way to re-trigger a build.

[dhui]

Nice, thanks for the refactors!