`newSuspendedTransaction` connecting to old database instances across test cases

Hi,
I'm working on a web service using Ktor 1.6.8 and Exposed 0.39.2.
My application module and database is setup as following:
```kt
fun Application.module(testing: Boolean = false) {
val hikariConfig =
HikariConfig().apply {
driverClassName = "org.postgresql.Driver"
jdbcUrl = environment.config.propertyOrNull("ktor.database.url")?.getString()
username = environment.config.propertyOrNull("ktor.database.username")?.getString()
password = environment.config.propertyOrNull("ktor.database.password")?.getString()
maximumPoolSize = 10
isAutoCommit = false
transactionIsolation = "TRANSACTION_REPEATABLE_READ"
validate()
}
val pool = HikariDataSource(hikariConfig)
val db = Database.connect(pool, {}, DatabaseConfig { useNestedTransactions = true })
}
```
I use ktor-server-test-host, Test Containers and junit 5 to test the service. My test looks similar like below:
```kt
@Testcontainers
class SampleApplicationTest{
companion object {
@Container
val postgreSQLContainer = PostgreSQLContainer(DockerImageName.parse("postgres:13.4-alpine")).apply {
withDatabaseName("database_test")
}
}

@Test
internal fun `should make request successfully`() {
withTestApplication({
(environment.config as MapApplicationConfig).apply {
put("ktor.database.url", postgreSQLContainer.jdbcUrl)
put("ktor.database.user", postgreSQLContainer.username)
put("ktor.database.password", postgreSQLContainer.password)
}
module(testing = true)
}) {
handleRequest(...)
}
}
}
```

I observed an issue that if I ran multiple test classes together, some requests ended up using old Exposed db instance that was setup in a previous test class, causing the test case failed because the underlying database was already stopped.
When I ran one test class at a time, all were running fine.
Please refer to the log below for the error stack trace:
```
2022-10-01 08:00:36.102 [DefaultDispatcher-worker-5 @request#103] WARN Exposed - Transaction attempt #1 failed: java.sql.SQLTransientConnectionException: HikariPool-4 - Connection is not available, request timed out after 30001ms.. Statement(s): INSERT INTO cards (...)
org.jetbrains.exposed.exceptions.ExposedSQLException: java.sql.SQLTransientConnectionException: HikariPool-4 - Connection is not available, request timed out after 30001ms.
at org.jetbrains.exposed.sql.statements.Statement.executeIn$exposed_core(Statement.kt:49)
at org.jetbrains.exposed.sql.Transaction.exec(Transaction.kt:143)
at org.jetbrains.exposed.sql.Transaction.exec(Transaction.kt:128)
at org.jetbrains.exposed.sql.statements.Statement.execute(Statement.kt:28)
at org.jetbrains.exposed.sql.QueriesKt.insert(Queries.kt:73)
at com.example.application.services.CardService$createCard$row$1.invokeSuspend(CardService.kt:53)
at org.jetbrains.exposed.sql.transactions.experimental.SuspendedKt$suspendedTransactionAsyncInternal$1.invokeSuspend(Suspended.kt:127)
at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:33)
at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:106)
at kotlinx.coroutines.internal.LimitedDispatcher.run(LimitedDispatcher.kt:42)
at kotlinx.coroutines.scheduling.TaskImpl.run(Tasks.kt:95)
at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:570)
at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:750)
at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:677)
at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:664)
Caused by: java.sql.SQLTransientConnectionException: HikariPool-4 - Connection is not available, request timed out after 30001ms.
at com.zaxxer.hikari.pool.HikariPool.createTimeoutException(HikariPool.java:695)
at com.zaxxer.hikari.pool.HikariPool.getConnection(HikariPool.java:197)
at com.zaxxer.hikari.pool.HikariPool.getConnection(HikariPool.java:162)
at com.zaxxer.hikari.HikariDataSource.getConnection(HikariDataSource.java:100)
at org.jetbrains.exposed.sql.Database$Companion$connect$3.invoke(Database.kt:142)
at org.jetbrains.exposed.sql.Database$Companion$connect$3.invoke(Database.kt:139)
at org.jetbrains.exposed.sql.Database$Companion$doConnect$3.invoke(Database.kt:127)
at org.jetbrains.exposed.sql.Database$Companion$doConnect$3.invoke(Database.kt:128)
at org.jetbrains.exposed.sql.transactions.ThreadLocalTransactionManager$ThreadLocalTransaction$connectionLazy$1.invoke(ThreadLocalTransactionManager.kt:69)
at org.jetbrains.exposed.sql.transactions.ThreadLocalTransactionManager$ThreadLocalTransaction$connectionLazy$1.invoke(ThreadLocalTransactionManager.kt:68)
at kotlin.UnsafeLazyImpl.getValue(Lazy.kt:81)
at org.jetbrains.exposed.sql.transactions.ThreadLocalTransactionManager$ThreadLocalTransaction.getConnection(ThreadLocalTransactionManager.kt:75)
at org.jetbrains.exposed.sql.Transaction.getConnection(Transaction.kt)
at org.jetbrains.exposed.sql.statements.InsertStatement.prepared(InsertStatement.kt:157)
at org.jetbrains.exposed.sql.statements.Statement.executeIn$exposed_core(Statement.kt:47)
... 19 common frames omitted
Caused by: org.postgresql.util.PSQLException: Connection to localhost:49544 refused. Check that the hostname and port are correct and that the postmaster is accepting TCP/IP connections.
at org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl(ConnectionFactoryImpl.java:303)
at org.postgresql.core.ConnectionFactory.openConnection(ConnectionFactory.java:51)
```

I have reported this issue to Ktor team at https://youtrack.jetbrains.com/issue/KTOR-4928 and found that this could be resolved by explicitly setting the db parameter for `newSuspendedTransaction` function.
So I assumed this could be a bug with `newSuspendedTransaction` when using implicit db. Please help to verify.

Thanks and regards,
Vinh.