SNOW-1675591 Fill in ExternalVolume and ExternalVolumeManager to do presigned url retrieval + blobname population

[sfc-gh-hmadan]

1. Add a GeneratePresignedUrls API call (new contracts + snowflakeClient API)
2. Enhance ExternalVolume.java to do presigned url file uploads to S3 (GCP and Azure testing pending thus commented out)
3. fix json field names where applicable
4. Add @JsonIgnoreProperties(ignoreUnknown = true) for forward compatiblity with service side evolution to some contracts that were missing this tag

[sfc-gh-psaha]

Looks mostly good. Just some minor comments.

[sfc-gh-psaha]

I feel like the background thread approach would also be simpler - you won't need all the Semaphore business. The bg thread could just keep refreshing in a loop whenever it sees the remaining urls hit a low.

[sfc-gh-hmadan]

Yes, couple reasons I kept this as simple as possible:

1. There can be many many external volumes (one per table), and the right fix there would be to have a small threadpool that multiplexes over all the active external volumes, and potentially even does a batched retrieval instead of one RPC per ExtVol. This is doable, just a matter of work ordering

2. rate matching - some ext vols will need 1 URL every minute, other volumes will need 10 URLs every second; to do a proper fix the low watermark needs to be adaptive instead of the hardcoded 5 right now.

3. Multipart upload support - to do multipart upload, each part requires its' own presigned URL AND needs to be associated with an upload-id; however now we can't cache URLs in advance as we won't know (in advance) how many parts any given file will have and don't have a batch of upload-ids either to quickly choose from.

Of these three, ended up just having a roundabout fix for (2) by allowing concurrent URL retrieval, as soon as there's high pressure on URL retrieval we'll end up caching a lot of URLs and reduce pressure.