[SPARK-17387][PYSPARK] Creating SparkContext() from python without spark-submit ignores user conf

## What changes were proposed in this pull request?

The root cause that we would ignore SparkConf when launching JVM is that SparkConf require JVM to be created first.  https://github.com/apache/spark/blob/master/python/pyspark/conf.py#L106
In this PR, I would defer the launching of JVM until SparkContext is created so that we can pass SparkConf to JVM correctly.

## How was this patch tested?

Use the example code in the description of SPARK-17387,
```
$ SPARK_HOME=$PWD PYTHONPATH=python:python/lib/py4j-0.10.3-src.zip python
Python 2.7.12 (default, Jul  1 2016, 15:12:24)
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from pyspark import SparkContext
>>> from pyspark import SparkConf
>>> conf = SparkConf().set("spark.driver.memory", "4g")
>>> sc = SparkContext(conf=conf)
```
And verify the spark.driver.memory is correctly picked up.

```
...op/ -Xmx4g org.apache.spark.deploy.SparkSubmit --conf spark.driver.memory=4g pyspark-shell
```

Author: Jeff Zhang <zjffdu@apache.org>

Closes #14959 from zjffdu/SPARK-17387.