public / sequelize

right now this is basically impossible to track down the root. this PR adds a more informative error message so you actually know what caused the error.

Before:

```
TypeError: sql.replace is not a function
        at Object.format (/Users/contra/Projects/staeco/node_modules/sequelize/lib/sql-string.js:87:14)
        at Object.format (/Users/contra/Projects/staeco/node_modules/sequelize/lib/utils.js:119:20)
        at Object.getWhereConditions (/Users/contra/Projects/staeco/node_modules/sequelize/lib/dialects/abstract/query-generator.js:2303:24)
```

After:

```js
Error: Invalid SQL string provided: ST_Intersects(123)
        at Object.format (/Users/contra/Projects/staeco/node_modules/sequelize/lib/sql-string.js:87:14)
        at Object.format (/Users/contra/Projects/staeco/node_modules/sequelize/lib/utils.js:119:20)
        at Object.getWhereConditions (/Users/contra/Projects/staeco/node_modules/sequelize/lib/dialects/abstract/query-generator.js:2303:24)
```

https://github.com/sequelize/sequelize/issues/6666

Makes it possible to pass condition objects (e.g. {foo: 'bar'}) to Sequelize.fn(), Sequelize.cast().
This makes it easy to use conditional counts in Sequelize queries, e.g.

Sequelize.fn('sum', {name: 'Fred'}) which results in SUM(`name` = 'Fred') which can be used in attribute definition
to count entities for which the condition matches.

* Failing testcase for findAll with nested include and inner join on m:n association

* Added insert queries to test

* Changed generateThroughJoin join identifier

* Improve code style

Association mixins refactoring

* Use soft versioning

for everything except inflection

* Update lodash

* Added upsertQuery for MSSQL

* Updated changelog and model api

* Add missing space

* Removed parenthesis from arrow functions

* Fixes edge case when two unique keys are present in model, added WITH(HOLDLOCK) to prevent concurrency issues, integration test for multiple unique keys

* Add ROW_FORMAT support to mysql query-generator

InnoDB's new Barracuda file format offers 2 new types of row formats
which offer significant improvements if working with utf8mb4 collation
by supporting index key prefixes up to 3072 bytes instead of the
default of 767 bytes offered with COMPACT or REDUNDANT row types
offered in the legacy Antelope file format.

This commit adds `ROW_FORMAT` support to the query-generator for
the mysql dialect by adding `options.format` for sequelize.define().

* Add tests for overrides global format option and inherits global format option

* Fix row-format test to use correct assert

* Add explicit test for options.format to SQL ROW_FORMAT

* Fix incorrect assert causing failing tests on row_format for options.format

* Correct changelog issue/hash to point to the same place.

* Rename options.format to options.rowFormat

* Fix missed format to rowFormat variable name

* fix(now): allow millisecons on createdAt and updatedAt

* test(timezone): timezone close check should allow 1 sec difference

* doc(changelog): add entry to change log

* test(now): create test for milliseconds in createdAt, updatedAt.

test(now): remove uneeded update code from test

removed unneeded file

* test(now): only run if mysql dialect

* test(lint): fix === lint error

* test(now): fix schema issue

* test(lint): remove space after comma

Not a valid built-in validator, throws an error when you try to use it

* add tests for postgres query-interface functions createSchema, databaseVersion, renameFunction
refactored file to address eslint warnings and errors

* Improve code style

https://greenkeeper.io/

* (fix) restore ignore field from deletedAt attr

* [ci skip] change log

https://greenkeeper.io/

* Update associations.md

* Update associations.md

* Update associations.md

* Update associations.md

* Update associations.md

Another correction

Fix es6 syntax, docs and changelog

* Small formatting fix

* Small formatting fix

https://greenkeeper.io/

* Fix unescaped field in MSSQL query generation

* Update changelog.md

* Update changelog.md