[CT-1892] [Bug] Non-standard abbreviations for kibibyte, etc

[dbeatty10]

Is this a new bug in dbt-bigquery?

• I believe this is a new bug in dbt-bigquery
• I have searched the existing issues, and I could not find an existing issue for this bug

Context

IEC 80000-13

Binary units of measurement express the size of data more accurately. When you compare the size of 100 KB to 100 KiB, the difference is relatively small, 2.35%. However, this difference grows as the size of the data values increases. When you compare the size of 100 TB to 100 TiB, the difference is 9.06% [1].

Current Behavior

Current code:

dbt-bigquery/dbt/adapters/bigquery/connections.py

Lines 258 to 269 in 3ce88d7

	def format_bytes(self, num_bytes):
	if num_bytes:
	for unit in ["Bytes", "KB", "MB", "GB", "TB", "PB"]:
	if abs(num_bytes) < 1024.0:
	return f"{num_bytes:3.1f} {unit}"
	num_bytes /= 1024.0
	num_bytes *= 1024.0
	return f"{num_bytes:3.1f} {unit}"
	else:
	return num_bytes

Expected Behavior

If we use logic like num_bytes /= 1024.0, then I'd expect it to align with IEC 80000-13 and use the following standardized abbreviations instead:

for unit in ["Bytes", "KiB", "MiB", "GiB", "TiB", "PiB"]:

(See the chart here for where I got the official abbreviations from.)

Alternatively, the current abbreviations for decimal multi-byte units could be kept and these substitutions could be made instead:

num_bytes /= 1000.0

and:

num_bytes *= 1000.0

Steps To Reproduce

Didn't actually reproduce -- just read the code!

Relevant log output

No response

Environment

- OS: N/A
- Python: N/A
- dbt-core: >= 1.0.0
- dbt-bigquery: >= 1.0.0

Additional Context

label: tech_debt

Today I Learned (TIL), format_bytes and format_rows_number are defined in dbt-core!
https://github.com/dbt-labs/dbt-core/blob/7b464b8a4957ec7969f19234020e110be1987923/core/dbt/utils.py#L456-L473

Context for the duplication

Removing them from dbt-core was mentioned here and here as part of the discussion for #48.

It feels to me like one of the following should have happened along with #48:

1. a simultaneous PR should have removed format_bytes and format_rows_number from dbt-core; or
2. #48 should have been added to dbt-core instead of dbt-bigquery.

As it stands, the implementations in dbt-core are unused (to the best that we can tell), but would probably need to be maintained to preserve backwards compatibility in case some unknown adapter is using them.

Proposal

Going forward, I'd propose that we choose one (and only one) of the following:

1. Keep separate implementations in dbt-core and dbt-bigquery (and fix the KB vs. KiB issue in both)
2. Remove the implementations in dbt-core (so they exist only in dbt-bigquery)
3. Completely move the implementation in dbt-bigquery back into dbt-core