FINERACT-2717: Check all currency usages before allowing currency removal

[Samer-Melhem-FOO]

Description

CurrencyWritePlatformServiceJpaRepositoryImpl#updateAllowedCurrencies allows removing a currency from the organisation's allowed-currency list. The existing in-use check only looked at loan products, savings products, and charges, so a currency still referenced by actual loans, savings/share accounts, client transactions, account transfers, cashier transactions, or GL journal entries could be silently removed.

This broadens the check to a JdbcTemplate-based scan across all relevant tables (m_product_loan, m_savings_product, m_share_product, m_charge, m_loan, m_savings_account, m_share_account, m_client_transaction, m_account_transfer_transaction, m_cashier_transactions, acc_gl_journal_entry) and improves CurrencyInUseException's message to be clearer to end users.

JIRA

https://issues.apache.org/jira/browse/FINERACT-2717

Test plan

• ./gradlew :fineract-core:compileJava :fineract-provider:compileJava passes
• Added CurrencyWritePlatformServiceJpaRepositoryImplTest (no prior test existed for this class), covering:
  • currency removed successfully when unused
  • CurrencyInUseException thrown when the removed currency is still referenced
  • the usage check is skipped entirely for currencies that remain allowed
• ./gradlew :fineract-provider:test --tests "...CurrencyWritePlatformServiceJpaRepositoryImplTest" — all 3 tests pass

[Samer-Melhem-FOO]

Two checks are currently failing here, both unrelated to this PR's change (CurrencyWritePlatformServiceJpaRepositoryImpl, CurrencyInUseException, OrganisationMonetaryConfiguration, plus a new unit test — no Liquibase/schema changes):

• run-integrations-tests-on-mariadb / MariaDB Tests (Shard 4 of 15): fails at jibDockerBuild with java.net.SocketException: Network is unreachable pulling azul/zulu-openjdk-alpine — the same transient Docker Hub registry issue seen on other PRs recently.
• run-integrations-tests-on-mysql / MySQL Tests (Shard 5 of 15): the app container built and started fine (Spring context came up, DB connection pool started), but the tenant Liquibase migration stalled for ~22 minutes with no progress between two log lines, and the health-check retry loop eventually timed out. This PR doesn't touch any Liquibase changelogs, so this looks like CI-runner resource contention/slowness rather than something caused by the change.

Could a maintainer re-run the failed jobs? Thanks!