[SPARK-31984][SQL] Make micros rebasing functions via local timestamps pure

[MaxGekk]

What changes were proposed in this pull request?

1. Set the given time zone as the first parameter of RebaseDateTime.rebaseJulianToGregorianMicros() and rebaseGregorianToJulianMicros() to Java 7 GregorianCalendar.

    val cal = new Calendar.Builder()
      // `gregory` is a hybrid calendar that supports both the Julian and Gregorian calendar systems
      .setCalendarType("gregory")
    ...
      .setTimeZone(tz)
      .build()

This makes the instance of the calendar independent from the default JVM time zone.

2. Change type of the first parameter from ZoneId to TimeZone. This allows to avoid unnecessary conversion from TimeZone to ZoneId, for example in

  def rebaseJulianToGregorianMicros(micros: Long): Long = {
    ...
      if (rebaseRecord == null || micros < rebaseRecord.switches(0)) {
        rebaseJulianToGregorianMicros(timeZone.toZoneId, micros)

and back to TimeZone inside of rebaseJulianToGregorianMicros(zoneId: ZoneId, ...)

3. Modify tests in RebaseDateTimeSuite, and set the default JVM time zone only for functions that depend on it.

Why are the changes needed?

1. Ignoring passed parameter and using a global variable is bad practice.
2. Dependency from the global state doesn't allow to run the functions in parallel otherwise there is non-zero probability that the functions may return wrong result if the default JVM is changed during their execution.
3. This open opportunity for parallelisation of JSON files generation gregorian-julian-rebase-micros.json and julian-gregorian-rebase-micros.json. Currently, the tests generate 'gregorian-julian-rebase-micros.json' and generate 'julian-gregorian-rebase-micros.json' generate the JSON files by iterating over all time zones sequentially w/ step of 1 week. Due to the large step, we can miss some spikes in diffs between 2 calendars (Java 8 Gregorian and Java 7 hybrid calendars) as the PR [SPARK-31959][SQL][test-java11] Fix Gregorian-Julian micros rebasing while switching standard time zone offset #28787 fixed and [SPARK-31986][SQL] Fix Julian-Gregorian micros rebasing of overlapping local timestamps #28816 should fix.

Does this PR introduce any user-facing change?

No

How was this patch tested?

By running existing tests from RebaseDateTimeSuite.

[SparkQA]

Test build #123982 has finished for PR 28824 at commit c5306ce.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[MaxGekk]

@cloud-fan @HyukjinKwon Please, review this PR.

[HyukjinKwon]

LGTM

[SparkQA]

Test build #124061 has finished for PR 28824 at commit b42d105.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[cloud-fan]

does this rely on JVM timezone?

[MaxGekk]

No, they don't. Let me move them out of the default JVM time zone. Thanks.

[SparkQA]

Test build #124103 has finished for PR 28824 at commit 381e950.

• This patch fails due to an unknown error code, -9.
• This patch merges cleanly.
• This patch adds no public classes.

[SparkQA]

Test build #124105 has finished for PR 28824 at commit 9f4f286.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[cloud-fan]

thanks, merging to master!