fix(e6): default single-argument LOG to natural-log base e

[tkaunlaky-e6]

What

DBR (and Spark/Hive/BigQuery) treat a single-argument LOG(x) as the natural logarithm (base e). e6 requires an explicit base, so a single-arg LOG currently transpiles to LOG(a), which fails at execution.

This supplies the default base e = 2.718281828459045 (fetched from DBR's e()) whenever a single-argument LOG is encountered:

SELECT log(a)     -->  SELECT LOG(2.718281828459045, a)

How

In the e6 generator:

• exp.Log single-arg → LOG(2.718281828459045, x) (covers e6/presto/snowflake/postgres read paths, which produce a Log node).
• exp.Ln where the source function name was log → LOG(2.718281828459045, x). This is the DBR path: Databricks parses single-arg log(a) into a natural-log (Ln) node tagged with the original name in meta. A genuine LN(x) is left as LN(x).

Two-argument LOG(base, value) is unchanged.

Verified

input (read=databricks)	output (e6)
log(a)	LOG(2.718281828459045, a)
log(a, b)	LOG(a, b)
ln(a)	LN(a)
log10(a)	LOG(10, a)

Tests

Added cases to test_math in tests/dialects/test_e6.py (DBR single-arg, two-arg, e6 identity). Full e6 + databricks suites pass, no regressions.