## What changes were proposed in this pull request?
1,create a libsvm-type dataset for lda: `data/mllib/sample_lda_libsvm_data.txt`
2,add python example
3,directly read the datafile in examples
4,BTW, change to `SparkSession` in `aft_survival_regression.py`

## How was this patch tested?
manual tests
`./bin/spark-submit examples/src/main/python/ml/lda_example.py`

Author: Zheng RuiFeng <ruifengz@foxmail.com>

Closes #12927 from zhengruifeng/lda_pe.