Concessão de memória não utilizada ao excluir

Eu tenho esta declaração de exclusão

delete from dbo.HighSchoolGradeLevel where Id=@P0

Este é o plano de execução

https://www.brentozar.com/pastetheplan/?id=H1qzvl3nW

quando o aplicativo está ocupado e há muitas exclusões chegando, o desempenho fica tão ruim, minhas perguntas são.

Como fazer com que essa exclusão tenha um desempenho melhor?

por que a memória concedida é tão alta, ela rouba a memória do buffer pool?

Nota: há uma cascata de exclusão nas 2 tabelas Highschoolcourse e highschoolgrade.

CREATE TABLE [dbo].[HighSchoolCourseGrade](
    [Id] [int] IDENTITY(1,1) NOT FOR REPLICATION NOT NULL,
    [TermSubTypeLkupId] [int] NULL,
    [Grade] [varchar](50) NULL,
    [GradePoint] [float] NULL,
    [CourseId] [int] NOT NULL,
    [CreatedDate] [datetime] NOT NULL,
    [UpdatedDate] [datetime] NOT NULL,
    [CourseUnits] [float] NULL,
 CONSTRAINT [PK_HighSchoolCourseGrade] PRIMARY KEY CLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]

GO

SET ANSI_PADDING ON

GO

/****** Object:  Index [ix_HighSchoolCourseGrade_CourseId_includes]    Script Date: 10/11/2017 4:33:38 PM ******/
CREATE NONCLUSTERED INDEX [ix_HighSchoolCourseGrade_CourseId_includes] ON [dbo].[HighSchoolCourseGrade]
(
    [CourseId] ASC
)
INCLUDE (   [Id],
    [TermSubTypeLkupId],
    [Grade],
    [GradePoint],
    [CreatedDate],
    [UpdatedDate],
    [CourseUnits]) WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
GO

ALTER TABLE [dbo].[HighSchoolCourseGrade]  WITH CHECK ADD  CONSTRAINT [FK_HighSchoolCourseGrade_CourseId] FOREIGN KEY([CourseId])
REFERENCES [dbo].[HighSchoolCourse] ([Id])
ON UPDATE CASCADE
ON DELETE CASCADE
GO

ALTER TABLE [dbo].[HighSchoolCourseGrade] CHECK CONSTRAINT [FK_HighSchoolCourseGrade_CourseId]
GO


CREATE TABLE [dbo].[HighSchoolCourse](
    [Id] [int] IDENTITY(1,1) NOT FOR REPLICATION NOT NULL,
    [CourseName] [varchar](200) NULL,
    [CourseTypeLkupId] [int] NULL,
    [EntryTypeLkupId] [int] NULL,
    [SubjectKey] [int] NULL,
    [GradeLevelId] [int] NOT NULL,
    [CreatedDate] [datetime] NOT NULL,
    [UpdatedDate] [datetime] NOT NULL,
    [TotalCourseUnit] [float] NULL,
    [TotalGradePoint] [float] NULL,
 CONSTRAINT [PK_HighSchoolCourse] PRIMARY KEY CLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]

GO

SET ANSI_PADDING ON

GO

/****** Object:  Index [ix_HighSchoolCourse_GradeLevelId_includes]    Script Date: 10/11/2017 4:32:58 PM ******/
CREATE NONCLUSTERED INDEX [ix_HighSchoolCourse_GradeLevelId_includes] ON [dbo].[HighSchoolCourse]
(
    [GradeLevelId] ASC,
    [SubjectKey] ASC
)
INCLUDE (   [Id],
    [CourseName],
    [CourseTypeLkupId],
    [EntryTypeLkupId],
    [CreatedDate],
    [UpdatedDate],
    [TotalCourseUnit],
    [TotalGradePoint]) WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
GO

ALTER TABLE [dbo].[HighSchoolCourse]  WITH CHECK ADD  CONSTRAINT [FK_HighSchoolCourse_GradeLevelId] FOREIGN KEY([GradeLevelId])
REFERENCES [dbo].[HighSchoolGradeLevel] ([Id])
ON UPDATE CASCADE
ON DELETE CASCADE
GO

ALTER TABLE [dbo].[HighSchoolCourse] CHECK CONSTRAINT [FK_HighSchoolCourse_GradeLevelId]
GO

Isso é de sp_blitzcahce

Atualizar

Encontrei este Delete no Query Store, eram 3 planos e forcei o SQL a usar um plano diferente. Estou me perguntando por que o SQL opta por usar o plano de execução mais caro?