Cache clauses -> row group mapping

Create a shadow table. For `stats`, it'd be `_stats_rowgroups`.

It contains three columns:

- the clause (eg `city = 'Dawson Creek'`)
- the initial estimate, as a bitmap of rowgroups based on stats
- the actual observed rowgroups, as a bitmap

This papers over poorly sorted parquet files, at the cost of some disk
space. It makes interactive queries much more natural -- drilldown style
queries are much faster, as they can leverage work done by previous
queries.

eg 'SELECT * FROM stats WHERE city = 'Dawson Creek' and question_id >= 1935 and question_id <= 1940`
takes ~584ms on first run, but 9ms on subsequent runs.

We only create entries when the estimates don't match the actual
results.

Fixes #6

parquet/parquet_table.cc

@@ -2,9 +2,7 @@
 
 #include "parquet/api/reader.h"
 
-ParquetTable::ParquetTable(std::string file) {
-  this->file = file;
-
+ParquetTable::ParquetTable(std::string file, std::string tableName): file(file), tableName(tableName) {
   std::unique_ptr<parquet::ParquetFileReader> reader = parquet::ParquetFileReader::OpenFile(file.data());
   metadata = reader->metadata();
 }
@@ -138,3 +136,6 @@ std::string ParquetTable::CreateStatement() {
 }
 
 std::shared_ptr<parquet::FileMetaData> ParquetTable::getMetadata() { return metadata; }
+
+const std::string& ParquetTable::getFile() { return file; }
+const std::string& ParquetTable::getTableName() { return tableName; }