datax mysql reader split too many tasks ,cause OOM

### mysql reader json configure 8 connections ,per connetion configure 1 table ;
channel configure to 16 , fit code condition eachTableShouldSplittedNumber = eachTableShouldSplittedNumber * 5;
when mysql table primary key use snowflake,data has billion and id has big range , the code can make huge size task, cause OOM; mysql table data size 100 million can make 1 million task
` int eachTableShouldSplittedNumber = -1;
if (isTableMode) {
// adviceNumber这里是channel数量大小, 即datax并发task数量
// eachTableShouldSplittedNumber是单表应该切分的份数, 向上取整可能和adviceNumber没有比例关系了已经
eachTableShouldSplittedNumber = calculateEachTableShouldSplittedNumber(
adviceNumber, originalSliceConfig.getInt(Constant.TABLE_NUMBER_MARK));
}

String column = originalSliceConfig.getString(Key.COLUMN);
String where = originalSliceConfig.getString(Key.WHERE, null);

List conns = originalSliceConfig.getList(Constant.CONN_MARK, Object.class);

List splittedConfigs = new ArrayList();

for (int i = 0, len = conns.size(); i < len; i++) {
Configuration sliceConfig = originalSliceConfig.clone();

Configuration connConf = Configuration.from(conns.get(i).toString());
String jdbcUrl = connConf.getString(Key.JDBC_URL);
sliceConfig.set(Key.JDBC_URL, jdbcUrl);

// 抽取 jdbcUrl 中的 ip/port 进行资源使用的打标，以提供给 core 做有意义的 shuffle 操作
sliceConfig.set(CommonConstant.LOAD_BALANCE_RESOURCE_MARK, DataBaseType.parseIpFromJdbcUrl(jdbcUrl));

sliceConfig.remove(Constant.CONN_MARK);

Configuration tempSlice;

// 说明是配置的 table 方式
if (isTableMode) {
// 已在之前进行了扩展和`处理，可以直接使用
List tables = connConf.getList(Key.TABLE, String.class);

Validate.isTrue(null != tables && !tables.isEmpty(), "您读取数据库表配置错误.");

String splitPk = originalSliceConfig.getString(Key.SPLIT_PK, null);

//最终切分份数不一定等于 eachTableShouldSplittedNumber
boolean needSplitTable = eachTableShouldSplittedNumber > 1
&& StringUtils.isNotBlank(splitPk);
if (needSplitTable) {
if (tables.size() == 1) {
//原来:如果是单表的，主键切分num=num*2+1
// splitPk is null这类的情况的数据量本身就比真实数据量少很多, 和channel大小比率关系时，不建议考虑
//eachTableShouldSplittedNumber = eachTableShouldSplittedNumber * 2 + 1;// 不应该加1导致长尾

//考虑其他比率数字?(splitPk is null, 忽略此长尾)
eachTableShouldSplittedNumber = eachTableShouldSplittedNumber * 5;
}
// 尝试对每个表，切分为eachTableShouldSplittedNumber 份
for (String table : tables) {
tempSlice = sliceConfig.clone();
tempSlice.set(Key.TABLE, table);

List splittedSlices = SingleTableSplitUtil
.splitSingleTable(tempSlice, eachTableShouldSplittedNumber);

splittedConfigs.addAll(splittedSlices);
}
} else {
for (String table : tables) {
tempSlice = sliceConfig.clone();
tempSlice.set(Key.TABLE, table);
String queryColumn = HintUtil.buildQueryColumn(jdbcUrl, table, column);
tempSlice.set(Key.QUERY_SQL, SingleTableSplitUtil.buildQuerySql(queryColumn, table, where));
splittedConfigs.add(tempSlice);
}
}
} else {
// 说明是配置的 querySql 方式
List sqls = connConf.getList(Key.QUERY_SQL, String.class);

// TODO 是否check 配置为多条语句？？
for (String querySql : sqls) {
tempSlice = sliceConfig.clone();
tempSlice.set(Key.QUERY_SQL, querySql);
splittedConfigs.add(tempSlice);
}
}

}`

Start at the table-mode branch shown in the issue, especially calculateEachTableShouldSplittedNumber and SingleTableSplitUtil.splitSingleTable. Reproduce the MySQL reader configuration with a single table, a snowflake-style primary key, and a large key range; done should mean task splitting no longer creates an excessive number of slices that causes OOM.

Written by the indexing model from the issue text.