Set file.encoding to UTF-8 for SBT tests

$ LANG=C.UTF-8 build/sbt "sql/testOnly *.JDBCV2Suite -- -z non-ascii"
...
[info] - column name with non-ascii *** FAILED *** (2 seconds, 668 milliseconds)
[info]   "== Parsed Logical Plan ==
[info]   'Project [unresolvedalias('COUNT('?), None)]
[info]   +- 'UnresolvedRelation [h2, test, person], [], false
[info]
[info]   == Analyzed Logical Plan ==
[info]   count(?): bigint
[info]   Aggregate [count(?#x) AS count(?)#xL]
[info]   +- SubqueryAlias h2.test.person
[info]      +- RelationV2[?#x] test.person
[info]
[info]   == Optimized Logical Plan ==
[info]   Project [COUNT(U&"\540d")#xL AS count(?#x)#xL AS count(?)#xL]
[info]   +- RelationV2[COUNT(U&"\540d")#xL] test.person
[info]
[info]   == Physical Plan ==
[info]   *(1) Project [COUNT(U&"\540d")#xL AS count(?)#xL]
[info]   +- *(1) Scan org.apache.spark.sql.execution.datasources.v2.jdbc.JDBCScan$$anon$1@df881a4 [COUNT(U&"\540d")#xL] PushedAggregates: [COUNT(`?`)], PushedFilters: [], PushedGroupByColumns: [], ReadSchema: struct<COUNT(U&"\540d"):bigint>
[info]
[info]   " did not contain "PushedAggregates: [COUNT(`名`)]" (ExplainSuite.scala:66)