BigQuery Storage API对clustered table的row restriction是否支持成本削减？

我使用Scala调用BigQuery Storage API的Java客户端，读取一个包含4个聚簇字段（f1、f2、f3、f4）的聚簇表，创建该表的命令如下：

bq mk [...] --table --clustering_fields f1,f2,f3,f4 mytable mytableschema.json

表结构如下：

[
  {
    "name": "f1",
    "type": "STRING",
    "mode": "REQUIRED"
  },
  {
    "name": "f2",
    "type": "STRING",
    "mode": "NULLABLE"
  },
  {
    "name": "f3",
    "type": "STRING",
    "mode": "REQUIRED"
  },
  {
    "name": "f4",
    "type": "STRING",
    "mode": "REQUIRED"
  },

  // other fields...
]

执行常规SQL查询时：

SELECT * FROM dataset.mytable 
WHERE f1 IN ('a', 'b') AND f2 IS NULL AND f3 = 'x'

系统会正确统计仅处理符合(a, null, x)和(b, null, x)聚簇的数据量。

但使用Storage API设置相同的row restriction读取时，账单显示按全表大小计费，API预估也显示全表将被计费。API核心使用逻辑如下：

val options =
    TableReadOptions
      .newBuilder()
      .setRowRestriction("f1 IN ('a', 'b') AND f2 IS NULL AND f3 = 'x'")
      .build()

val readSessionBuilder =
  ReadSession
    .newBuilder()
    .setTable(tableName)
    .setDataFormat(DataFormat.AVRO)
    .setReadOptions(options)

val readSessionRequestBuilder =
  CreateReadSessionRequest
    .newBuilder()
    .setParent(ProjectName)
    .setReadSession(readSessionBuilder)
    .setMaxStreamCount(1)

val session = client.createReadSession(readSessionRequestBuilder.build())
// ... read from session.getStreamsList

请问BigQuery Storage API是否支持通过row restriction对聚簇表实现成本削减？我未找到相关说明文档。

内容的提问来源于stack exchange，提问作者Giovanni Caporaletti