Merge with ADBC 21

[jasonlin45]

pulls in a stupid amount of changes...

Tested with Jaffle Shop init - Snowflake and Bigquery are all working with no issues.

Databricks needs an FS PR.

How to Review

The main changes that we care about will be in the go/adbc folder.

There are a ton of changes that got pulled in, but the majority of these changes will not be our Go drivers that we use.

Many of these changes are small and the product of upgrades to the driver base and to Arrow.
e.g.

-r.currentBatch = r.stream.Record()
+r.currentBatch = r.stream.RecordBatch()

To fix this error, the check should parse any URL(s) found on the line and accept the line only if at least one of the parsed URLs has a hostname of exactly img.shields.io. Since the code processes lines from Markdown source files and looks for badge images (with links of the form ![tag](https://img.shields.io/...)), we should extract any URL(s) within parentheses following markdown image syntax, parse the URLs using Python's standard urllib.parse.urlparse, and check their .hostname field.

• Import urllib.parse.urlparse at the top.
• Replace the substring check with code to find all URLs in the line, parse each, and check if any have hostname img.shields.io.
• Only process the line accordingly if such a URL is found.
• Only edit the relevant lines in the shown code (we are limited to what code blocks we have).

To fix the issue, replace the insecure call to File.createTempFile with a secure version using the java.nio.file.Files.createTempFile API, which creates files with safe permissions (-rw------- on unix-like systems). We'll obtain a Path object from Files.createTempFile, then construct a File from it, to preserve usage in the rest of the code.
In file java/driver/jni/src/test/java/org/apache/arrow/adbc/driver/jni/JniDriverTest.java, line 138 should be replaced. Since the code uses File APIs later on, we will add an import for java.nio.file.Files and update the line to:

File tmpPath = Files.createTempFile("jni_test", ".sqlite").toFile();

No further code changes are necessary, since this preserves the existing File-based usage.

---

[serramatutu]

@jasonlin45 this is A LOT of files... Do you have a pointer on where to start reviewing this? Have you changed anything that came from ADBC upstream?

[jasonlin45]

@serramatutu I added a how to review to the PR description - most of the changes we need to review aren't super large within the drivers themselves

[chasewalden]

@serramatutu, I'd also assume that many of the changes outside of the go/adbc/... subtree can more-or-less be ignored. To my understanding, we have mostly focussed on the go codebase, but please correct me if I am wrong @jasonlin45.

I will start my review now, but it will likely take me a bit to get through it.

---

Side note: Is it worthwhile for us to try to automate regular syncs with upstream? That would (hopefully) minimize divergence and make future "catch-up" PRs not have 550+ commits

[xuliangs]

Thank you Jason, this is huge.